# SEO Master Playbook
## Compiled from 1,022 transcripts / ~290 hours of SEO courses, condensed into executable guidance for website-builder agents

**Coverage:** Google Search Central (official), Ahrefs Academy (SEO Beginners, Technical, Keyword Research, Advanced Link Building, AI SEO/AEO, Local, YouTube, Blogging, Content Marketing, Affiliate), Semrush courses (Essentials, Backlinks, New Era SEO, LLM Visibility, Local, SERP mastery), Moz (Whiteboard Friday, One-Hour Guide, Daily Fix), Income School, Matt Diggity, Neil Patel SEO Unlocked, Brian Dean/Backlinko, Yoast, HubSpot AEO + 47 mega-course videos (programmatic, ecommerce, AI search, technical audits, local).

> **How to read this doc.** Part 1 is the playbook, ordered, executable, deduplicated across sources. Part 2 is the per-course detail appendix (the raw distilled notes, including the full Ahrefs AEO course). Part 3 is the source index.
>
> **Source hierarchy when advice conflicts:** (1) Google official guidance > (2) cross-course consensus > (3) practitioner tactics > (4) grey/black-hat methods (flagged, never default). Third-party metrics (DR/DA/Authority Score) are proxies, not Google ranking factors. No tactic in this document guarantees any ranking.

> **AGENT BRIEF: read this first if you’re an AI agent.**
> 1. **Read Part 1 in full** (Ch.0–16, about 300 lines). Don’t load the whole file. Open Part 2 only for the chapter you’re executing (use the headings to find it).
> 2. **Route by site type** and skip what doesn’t apply:
>    - content or data site: Ch.0, 2–7, 12–16;
>    - local business: add Ch.9;
>    - shop: add Ch.10;
>    - video: add Ch.11 and the "YOUTUBE SEO" appendix.
>    - Programmatic or templated sites (many pages from one template) must pass the Ch.12 doorway test before anything else.
> 3. **Paid tools are optional.** Where a step names Ahrefs, Semrush, Brand Radar or Content Explorer, use the free equivalent: Google Search Console, Bing Webmaster Tools, Google and YouTube autosuggest, "People also ask", the live results page, Rich Results Test, validator.schema.org, and your own crawl.
> 4. **Never do these without the human’s explicit OK:**
>    - off-site actions: outreach emails, posting, buying links, creating profiles or citations;
>    - bulk noindex, redirect or consolidation of pages that already get traffic.
> 5. **Corrections to the body:**
>    - HowTo rich results were retired by Google in 2023. Mark up HowTo only as machine-readable structure, and don’t expect a rich result.
>    - FAQ rich results only show for authoritative government and health sites.
>    - Vendor statistics (correlations, % lifts) are directional claims, not facts.
>    - Ch.1’s "WordPress" advice doesn’t apply to custom-built sites.
> 6. **Every change must be verifiable.** Prefer automated checks (a test or audit script) over one-off fixes, so the rule can’t quietly regress.

---

# PART 1, THE PLAYBOOK

## Ch.0, The non-negotiables (cross-course consensus)

Every school disagrees on tactics; these are the points they converge on:

1. **Search intent is the first filter.** Match the SERP’s dominant intent (informational / navigational / commercial / transactional) and format before writing anything. A page that mismatches intent cannot rank, full stop.
2. **Cover topics, not keywords.** Topical authority = answering every relevant question in a niche, interlinked. Koray’s case: 0→300k+ monthly organic visits on new domains with zero links, purely from topical-map-complete publishing. Plan clusters, not posts.
3. **Real experience beats generated volume.** Post-HCU winners show first-hand signals: original photos, named authors, tests, data. AI assists outlines/drafts; humans own expertise.
4. **Crawlable + indexable + fast-enough is the technical floor.** Perfect Lighthouse scores are optional; broken links, orphan pages, accidental noindex are fatal. (Ecommerce distillation: only 4 technical things matter, indexation, no broken links, UX, hierarchy.)
5. **Money pages first.** Rank collection/category/product/comparison pages before informational content. Blog traffic that doesn’t convert is an expense. (Puzzle brand: $120k of blogs → $2k/mo; 40 collection pages → $10k+/mo in 4 months.)
6. **Links still matter; quality >> quantity.** 10–20 relevant links beat 1,000 junk ones. One relevant editorial link can outweigh millions of low-quality links (John Mueller). Vet every linking domain.
7. **Anchor text is a per-page science.** Penguin judges per URL. Reverse-engineer the top-3 anchor distribution per target page; never reuse identical anchors; brand/URL/generic anchors dominate natural profiles.
8. **Velocity must match stage.** Sandbox sites: trust links first (profiles, citations, 1–3 new-page links/week). Established sites: link volume justified by traffic. Too fast = stuck sandbox or penalty; too slow = never close the gap.
9. **Distribution is half the job.** Publishing without promotion = invisible. Internal links, email, social, communities, and digital PR, Google is not the only search surface anymore (Search Everywhere: YouTube, Reddit, Amazon, AI chats).
10. **SEO compounds; systems beat heroics.** Content audits, striking-distance reviews, refresh cadences, and GSC diagnosis are recurring workflows, not one-time projects.
11. **AI search is a new surface, not SEO’s replacement.** Same fundamentals apply; the new work is citation-worthiness, entity clarity, brand mentions, and presence where LLMs look (Reddit, YouTube, G2, comparison pages).
12. **Never ship pages you wouldn’t show a real visitor.** If it exists only for search engines (location-swap pages, hidden islands, template spam), it’s a penalty candidate, see the HVAC case (–63% traffic).

---

## Ch.1, New site launch checklist

**Before building:**
- [ ] Niche slicing exercise (Income School "brand plan"): split the topic into logical categories on paper → these become the site structure and content clusters.
- [ ] Pick max 3 subgroups to launch with. Depth in one lane > scattered coverage.
- [ ] Verify demand: Google Trends (topic-level, 5-yr) per subgroup; competitor top-10 analysis; confirm SERPs contain sites of your class (if page 1 is all DR80+, pivot the subgroup).
- [ ] Decide domain strategy: one domain unless audiences/messaging genuinely diverge (Moz: single domain concentrates authority; Google supports both subfolder/subdomain, pick for ops reasons, track separately if subdomain).
- [ ] WordPress or equivalent CMS for content-heavy sites (Yoast: a 10-yr-old static site only deserves rebuilding if you’ll actually keep updating it).

**Build-time essentials:**
- [ ] Semantic HTML: real `<a href>` links (never JS-only div "links"), History-API routing (no hash-router URLs), every route returns HTTP 200 when opened directly, content in initial HTML where practical (JS-rendered-only content indexes late or never).
- [ ] Clean hierarchy: every page reachable in ≤3 clicks; nav + footer + breadcrumbs; no orphan pages.
- [ ] Heading structure: one H1, logical H2/H3 (screen readers + Google’s parser both use it).
- [ ] Core Web Vitals on real devices (low-end phone + slow network test; lab scores ≠ field data).
- [ ] Mobile = the indexed version. Test every template on mobile.
- [ ] XML sitemap auto-generated + submitted; robots.txt permits the money paths; no stray `noindex`.
- [ ] 404 page with search + popular-links (QRG: plain 404 = medium quality, helpful 404 = high).
- [ ] Schema from day one: Organization/LocalBusiness, Breadcrumb, Article/Product/FAQ as applicable, complete, not partial.
- [ ] author/about/contact pages, entity clarity and trust surface (matters for Google E-E-A-T *and* LLM brand resolution).

**Launch-phase trust links (Diggity Stage-1):**
- [ ] Social profiles (all at once is fine, natural).
- [ ] ~30–40 business citations / directories over a few days (natural business behavior).
- [ ] Then 1–3 new-page links/week (guest posts, podcasts, mentions) spread across homepage + money + supporting pages.
- [ ] DO NOT burst links at a new domain, his 100-links-in-2-months site spiked then died.

---

## Ch.2, Keyword & topic research workflow

**Step 1, Seed pool (combine methods):**
- Brainstorm via customer shoes + AI expansion (Ahrefs: seed → 20 AI variants → Matching Terms report).
- Competitor content gap: their top-10 keywords where you’re absent from top-100; sort by their traffic (Ahrefs Content Gap / Semrush Keyword Gap).
- Competitor PAID keywords: CPC proves profitability, high CPC = high value ≈ high organic difficulty ($40M ecommerce method).
- Your own GSC/organic keywords → dropped rankings → related questions to rebuild coverage.
- Community mining: Reddit/Quora/forums for question language (Moz + Semrush both emphasize this for *search everywhere* intent).
- People Also Ask: dynamic, click a question and the suggestions shift → that’s Google showing its internal topical map (Koray). Mine PAA recursively.

**Step 2, Qualify:**
- Intent check on the live SERP (not the tool’s label): dominant intent + format wins.
- Intent-split check: "identify intents" tools show traffic distribution across intents, one page can serve several.
- Difficulty reality: KD ≤10–20 for new sites; SV≥500 optional, **traffic potential of the parent topic matters more than single-keyword volume** (one page ranks for thousands of variants).
- Striking distance: positions 4–20 on existing pages = fastest wins (Moz), always check before creating new pages.
- CPC as value proxy when volumes are vague (Google Keyword Planner ranges are wide, "10k–100k", treat all volumes as directional).

**Step 3, Prioritize into clusters:**
- Cluster by shared entities/attributes (ontology → internal-link map later).
- Order: money/BOFU first → supporting commercial → informational that feeds the cluster.
- Include zero-volume entities that complete the topical map (unique-source advantage per Koray; but prune to correct context, "apple" fruit vs company).
- Income School alternative for low-competition niches: response posts across the grid → Battleship analysis → upgrade only proven hits.

---

## Ch.3, Content operations

**The topical map system (highest-leverage content strategy in the corpus):**
1. Define the core entity + attribute tree; list every question per attribute.
2. Write briefs per question with required entities/sub-questions; granularity scales with competition.
3. Publish bottom-up: BOFU cluster first (earlier Moz/Income School point, faster revenue proof).
4. Cover the map completely, partial coverage reads as "not the authority".
5. Writers: give them semantic briefs (entities to mention, questions to answer), Koray found law/philosophy/engineering backgrounds execute these best.

**Battleship method (portfolio management):**
- Phase 1 Grid: fast response posts (~1 hr, ~1,300 words) covering the niche map.
- Phase 2 Analyze: classify posts, hit / upgrade / miss (never touch again) / maintainer.
- Phase 3 Invest: pillar-level effort only on proven winners (a #4 response post justifies the 8-hour deep version).

**Quality bar (E-E-A-T, post-HCU):**
- First-hand proof: original images/tests/data; named authors w/ Person schema + knowsAbout; methodology shown.
- AI use = outlines, drafts, repurposing; human expertise = the differentiator (Moz + every 2025 source).
- Unique value per page: if a page adds nothing the SERP doesn’t already have, don’t publish it.

**Content audits (Semrush/Moz/Income School merge):**
- Quarterly: classify every page → keep / upgrade / consolidate / redirect / prune.
- Consolidate cannibalizing pages (same intent, split signals) → one stronger URL + 301s.
- Page-2/3 content with potential → refresh + internal links (Moz striking-distance).
- Update top performers ≥ annually (Moz); prune dead weight, Semrush’s audits found UX/conversions improve after removing low-quality content.

**Featured-snippet engineering (Income School, do on every post):**
- Write a 2–3 sentence **answer target**: standalone (makes sense without the question), starts with the noun (not "it/they"), encyclopedia tone (semi-formal, no slang/opinion), preceded by a tee-up sentence, **bolded**.
- 4 snippet types: paragraph (~50%), list, video, table (underserved, build table snippets for comparisons deliberately).

---

## Ch.4, On-page rules

- **Keyword placement (Edward Show rule):** keyword in title, meta description, URL slug, H1, first sentence, then stop. No stuffing variations through H2s/body.
- **Titles:** descriptive + distinguishing attributes (ecommerce: brand + color + key spec); front-load value; compare against what’s actually showing in the SERP, Google rewrites bad titles.
- **Meta descriptions:** CTR lever, not ranking. For AI search, **meta matters MORE** (Mike King via HubSpot): LLMs want the answer spoiled, put the direct answer in the meta.
- **Headings:** logical H1→H2→H3 outline that could stand alone as a summary.
- **URLs:** stable, descriptive; never change casually; extensions don’t matter (.html vs clean, Google).
- **Images/media:** real filenames, alt text, original > stock; video captions for accessibility + parseability.
- **Word count follows SERP intent**, not doctrine, check the median ranking page; comprehensiveness ≠ bloat.
- **Write natural phrases:** "HVAC repair in Los Angeles," never fragment-H1 "HVAC repair Los Angeles."

---

## Ch.5, Architecture & internal linking

- Structure models (pick deliberately, document it): **silo** (strict topic compartments, max topical clarity), **hub-and-spoke** (hub ↔ spokes bidirectional), **content library**, **topic gateway**, **database/glossary** for scale.
- Interlinking rule (Koray’s ontology): link where two pages share an entity/attribute, the mutual point IS the link point.
- Equity routing: identify your most-linked pages (Ahrefs Best-by-Links / Semrush Indexed Pages) → add contextual internal links from them to money pages.
- Every money page gets supporting content that links to it (Income School staple/pillar layer; Diggity: risky-to-direct-link pages get fed via internal links).
- Prevent cannibalization by design: map focus keyword → single URL in the content map before creating; Rank Math/SEO plugins flag collisions, fix by merging, not adding.
- Hub pages for programmatic/comparison fleets, visible in nav, never islands.
- Breadcrumbs on all content pages (+ Breadcrumb schema).

---

## Ch.6, Technical SEO checklist

**Crawl & index:**
- [ ] robots.txt: blocks waste (facets, internals), never money paths. Google ignores `crawl-delay`, rate adjusts automatically to server health. robots.txt controls crawling, NOT indexing (noindex for that, and never noindex pages you want rendered via JS).
- [ ] XML sitemap(s): split by type under an index, truthful lastmod (fake freshness = corroborated and ignored, Google checks ETags/Last-Modified/structured dates).
- [ ] Indexation debugging order: GSC Pages → URL Inspection (fresh data) → crawled? (robots/removals/server errors) → noindex? → canonical pointing away? → fix → validate + Request Indexing.
- [ ] Orphan pages: zero internal links in = dead pages. Fix all.
- [ ] Canonicals = hints, not commands; they’re NOT redirects. Don’t canonicalize materially different pages (out-of-stock vs replacement) to hoard equity.
- [ ] Duplicate content: SiteLiner >20% match on money pages → canonical (legit dupes: print versions, variants) or 301 (thin dupes).
- [ ] Redirects on every URL change, 301 for permanent; during migrations change ONE thing at a time (never domain + URLs + server + content + robots simultaneously, Google’s migration guidance).

**JavaScript SEO (Google’s own series, critical for modern builders):**
- Content rendered only client-side indexes late or never → put critical content in initial HTML; SSR/SSG/hybrid preferred over client-only rendering.
- Normal links (`<a href>`), Google can’t follow click-handlers; History API routes with real URLs, not `#/fragments`.
- Each route must serve HTTP 200 standalone + render correctly direct-loaded.
- Never ship `noindex` in initial HTML planning to remove it later via JS, that JS may never execute for Googlebot.
- Dynamic rendering = legacy fallback only, not a strategy.
- Test: URL Inspection "Test Live URL" + rendered screenshot, Rich Results Test, incognito.

**Performance/experience:**
- CWV is field data (CrUX), lab Lighthouse scores are diagnostic only. Test on low-end phones + throttled networks.
- Cache CSS/JS/API responses Google must fetch; avoid POST-based resource patterns that can’t cache.
- AMP = not a ranking factor.

**WP/CMS-specific traps (Yoast/Rank Math):**
- "Discourage search engines" checkbox = sitewide noindex, verify on every launch/migration.
- Theme switches wipe footer/social links, re-check after every theme change.
- Tag archives = thin auto-pages; noindex thin archives; categories broad + curated.
- Canonicals in plugin Advanced settings can silently mark pages as duplicates, audit after imports.

---

## Ch.7, Structured data & rich results

- Schema’s job now: rich snippets **and** machine-parsability for AI surfaces.
- Match type to page purpose, Article, Product (+offers/reviews/price), LocalBusiness (+NAP, openingHours), Course, Book, FAQPage, HowTo, Video, Breadcrumb, Person (author + knowsAbout), Organization (+sameAs).
- Fill completely, keep truthful (review/schema abuse = manual action).
- FAQ schema only where real visible Q&A exists (Google now restricts FAQ rich results to authoritative sites, still useful as machine-readable structure).
- Validate: Rich Results Test on templates before shipping.

---

## Ch.8, Link building & digital PR

**Prospect vetting (Semrush + Diggity + Gotch):**
- [ ] Indexed? (site: search), unindexed domain = worthless or toxic.
- [ ] Wayback history, same niche/owner? Pivoted foreign-content sites = red flag.
- [ ] Server neighborhood, shared IP surrounded by casino/pharma/adult = network signal.
- [ ] Organic traffic exists + topical relevance, a DR number alone means nothing.
- [ ] Footprint check, identical templates/stock images across "different" sites = PBN.
- [ ] Avoid domains visibly hit by quality updates or showing bot traffic curves.

**Acquisition tactics by risk tier:**
- **Core (always safe):** linkable assets (original data/research, free tools, ChatGPT builds calculators in ~1hr, epic guides, custom design, rankings, steal *angles* from other industries: Gotch’s blue-ocean linkbait), digital PR/journalist pitching (homepage links feed the whole site), broken-link building + Moving Man (dead/rebranded companies' links → notify + pitch replacement), guestographics (infographic + offer unique intro text, removes the publisher’s work barrier), resource pages, unlinked-mention reclamation, competitor link-gap/intersect.
- **Judgment calls:** link insertions on already-trafficked pages ($100–400; beats guest posts on traffic guarantee, paid placements are a spam-policy gray area; vet the host like above), guest posts on real sites with real traffic, 3-way exchanges, directory/citation volume.
- **Flagged (don’t default):** PBNs, tiered link schemes, mass exchange networks, the Merger technique (acquire-site→301, it works but is textbook manipulation), buying links without vetting. Diggity material documents these deeply; Google’s spam policies don’t care that practitioners get away with them.
- **Income School’s alternative:** in weak niches, skip link building entirely, hard-to-replicate content (original experiments, unique data, own photos) earns links passively (their cookforfolks: 400+ RDs, zero outreach). Verdict: right for hobby/info sites, wrong for competitive commercial SERPs where deliberate acquisition is required.

**Anchor text (per-page discipline, Diggity webinar):**
- Penguin evaluates per URL. For each money page: pull top-3 competitors' anchor profiles → classify (target/partial/brand/URL/generic/misc) → average = the niche’s natural distribution → build toward it.
- Exclude mega-authority outliers ranking on 2 links (they skew ratios).
- Never reuse identical target anchors; vary phrasing; exact-match ≤2× total; SEO-title-tag anchors are the most natural type (forums/CMSs default to them).
- If everyone on page 1 ranks with 3–5 links, don’t be the page with 20 (black-sheep effect, you can’t out-link an authority gap).
- Traffic must justify velocity (every real linker visited the site once).

**Gap math:** bucket competitor RDs by authority tier, average, subtract yours, scale by exponential DR factor, exclude nofollow.

---

## Ch.9, Local SEO

- **GBP first:** primary category + all relevant secondary categories; every service listed; complete profile; keyword-rich description; photos (real); products/services filled.
- **Ranking trio:** relevance × distance × prominence. You can’t change distance; dominate the other two.
- **Reviews:** steady velocity > total count; ask systematically (post-service automations); respond to all incl. negatives; keywords in responses help relevance; never buy/gate.
- **Citations/NAP:** consistent everywhere; audit once, maintain on change; Yoast + all local courses treat as table stakes.
- **Location pages (the doorway trap, read this before generating):** unique per-location content (local details, staff, reviews, service differences) or don’t build them. The HVAC case: hundreds of swapped-name pages → –63% organic after the March 2024 core update; recovery only via consolidated genuinely-local pages.
- Local schema (LocalBusiness + geo + sameAs), embedded map, localized content sections.
- Local links: sponsorships, local press, community orgs, local relevance beats raw metrics.
- Service pages rank by service, not by fake city-pages, one strong service page + GBP + reviews > 30 doorway pages.

## Ch.10, Ecommerce SEO

- **Money pages = collections/categories + products.** Expansion lever: collection pages per real modifier combo (color/size/material/style/audience/occasion), with genuinely different merchandising.
- Product pages: descriptive titles w/ distinguishing attributes (brand/color/spec), unique descriptions (not manufacturer copy, duplicate across resellers), product schema complete (price/availability/reviews), review systems on.
- Out-of-stock: keep URL + offer alternatives (users still arrive); don’t canonicalize to a different product; redirect only if truly replaced.
- Cover the shopping journey: homepage → category → product → reviews → buying guides/comparisons. Buying guides feed collection pages via internal links.
- Link tactics: insertions on trafficked pages, digital PR to homepage, journalist relationships, 10–20 quality links move most mid-size stores.
- KPI = organic revenue + top-3 on transactional terms. Traffic without revenue = the agency trap ($120k puzzle-brand lesson).

## Ch.11, YouTube & video SEO

(Condensed from Ahrefs YouTube course + Brian Dean, see appendix for depth.)
- Keyword research YouTube-side: autosuggest + competitor video keywords + tools; target terms where Google shows video carousels (dual-surface win).
- Retention is THE metric: hook in first 15s, pattern-interrupts, tight editing; CTR (thumbnail+title) × retention = distribution.
- Titles front-load keyword; descriptions carry the summary + chapters; chapters = Google "key moments" indexing.
- Video schema (VideoObject) + embed on matching pages, the page and the video rank each other.
- YouTube content feeds LLM visibility (transcripts are citable text), record answer-style segments.

## Ch.12, Programmatic SEO

- **Model:** head term + modifier ("{city} {service}", "{product} vs {product}", "{stat} calculator"). Head term must have real volume; each page individually weak → low-KD head terms only.
- **The doorway line (repeat, it’s the #1 pSEO failure):** if pages differ only by the variable, are hidden from nav, offer no unique value, or exist just to capture query variants → penalty. Google’s doorway abuse definition is explicit.
- **Right way:** variable dependencies, the variable must change the page’s actual content (data, comparisons, local specifics, computed outputs). Best pSEO pages are utilities (calculators/tools/databases), not text templates.
- Data: buy datasets (Bright Data/Coresignal), public data (census/SEC/court/WHOIS), or scrape + **transform** via LLM, never republish raw scrapes.
- Surface via visible hub pages in nav.
- **Indexing at scale is the bottleneck:** indexation rate tracks site authority, expect a fraction indexed on new domains; drive via sitemaps + internal links; monitor per-template index rate; prune templates that never earn traffic ("celebrity email" pages ranked but converted ~0 for B2B).
- Fragment warning applies doubly: template H1s must be natural phrases.

## Ch.13, AI search: AEO / GEO / LLM visibility

(Full course text preserved in Appendix, Ahrefs AEO course. This is the operational digest.)

- **How AI answers work:** query fan-out (one prompt → multiple background searches), retrieval → citation selection; each platform differs (AIO over-indexes on Google ranking; ChatGPT leans Bing/ecosystem sources; Perplexity is citation-forward).
- **Three visibility types:** linked citations, unlinked mentions, recommendation presence, spectrum not binary (Ahrefs framework).
- **What gets cited:** pages that directly answer the question with extractable structure (answer blocks, tables, stats), brand entities LLMs can resolve (consistent name/description everywhere), third-party validation (G2, Reddit, comparisons, press), meta descriptions that spoil the answer.
- **Page types that win AI traffic:** comparison/versus articles (the top driver, Ross Hudgens data), complex interactive tools/calculators (defensible vs AIOs), genuine experience content, unique data.
- **Vs-page format:** "us vs them" H1 + early CTA + feature-comparison table high on page; comparisons grounded in real sales-call objections; triopoly comparisons enter 3-way conversations as the 4th player.
- **Brand-mention tiers:** own channels → adjacent communities (Reddit—an own-subreddit engagement correlates with positive mentions across ALL of Reddit; YouTube) → earned third-party (reviews sites, press, expert quotes).
- **Don’t chase citation-share vanity metrics:** cross-reference citations with real traffic (AI-slop domains rack up citations with no influence; retrieval < citation).
- **Tech for AI:** GPTBot/robots decisions are deliberate choices (block = invisible to ChatGPT); llms.txt is emerging not official; structured data feeds machine understanding; fast+parseable sites get crawled more by AI agents.
- **Measurement:** GA4 regex for AI referrers (chatgpt.com|perplexity.ai|claude.ai|copilot); monthly manual prompt audit (10 money questions × ChatGPT/Perplexity/AIO, log cited sources); branded-search-volume growth in GSC as the demand proxy.
- **Reality check:** AI answers are still volatile and often wrong; AI traffic ≈ small % of total today but the *buyers it sends are late-funnel*, treat as high-value channel, not replacement for SEO.

## Ch.14, Measurement & troubleshooting (GSC playbook)

- **Performance report dimensions:** queries, pages, countries, devices, search appearance, date, analyze in combinations, never aggregate-only.
- **Diagnostic patterns:** high impressions + low CTR → title/snippet problem; expected queries missing → content/intent gap; important pages missing → indexation check; mobile vs desktop divergence → UX issue; country split → localization.
- Segment branded vs non-branded (regex excluding brand); navigational queries ("brand login") = trust signal.
- **Traffic-drop sequence:** compare like periods (16-mo view) → segment by page/query/device/country/search-type → check Google Trends (seasonality/demand shift, not penalty) → check indexing/manual actions → check announcement window vs core updates → check technical changes (deploy log).
- **Crawl stats:** host status/DNS/connectivity; crawl response mix; file types; purpose (refresh vs discovery); per-crawler breakdown. Crawl budget only matters at ~1M+ URLs, server instability is a server problem, not a budget problem.
- **Ranking reports for stakeholders:** organic revenue/conversions first, positions second, traffic third (Semrush + ecommerce consensus).

## Ch.15, Maintenance system

- Weekly: GSC errors + performance anomalies.
- Monthly: striking-distance report (positions 4–20) → targeted improvements; AI-visibility prompt audit; review GBP.
- Quarterly: content audit (keep/upgrade/consolidate/prune); link-profile vetting; technical crawl (Screaming Frog); refresh top pages.
- Annually: every top-10-traffic page reviewed and updated (Moz rule); competitor gap re-run; schema/coverage re-audit.
- After every deploy: spot-check rendered HTML, title/meta, canonicals, nav links, no accidental noindex, deploys are the #1 self-inflicted SEO incident.

---

## Ch.16, Anti-patterns (penalty triggers, consolidated DON’T list)

1. Location/keyword-swap page fleets (doorway abuse), HVAC: –63% traffic.
2. Anchor-text monotony, same exact-match anchor repeatedly across links.
3. Link bursts at new domains (Diggity beard-oil: spiked → declined → manual penalty).
4. Fiverr-style link packages / unvetted marketplace links / PBNs.
5. Content fleets with no unique value, scaled AI output without experience/data.
6. Fake `lastmod` dates to simulate freshness.
7. Canonicalizing materially different pages to hoard equity.
8. Changing domain + URLs + server + content in one migration.
9. `noindex`/`robots` mistakes on money paths; "Discourage search engines" left on.
10. JS-only links, hash routing, client-only content for critical paths.
11. Buying reviews / review-gating; spammy GBP categories.
12. Chasing impressions/traffic that can’t convert (blogs-for-traffic ecommerce trap).
13. Over-optimizing meta/Lighthouse scores while money pages languish (80/20).
14. Treating tool metrics (DR/DA) as Google ranking factors.
15. Guaranteeing rankings to stakeholders, nobody can.

---

---

# PART 2, PER-COURSE DETAIL APPENDIX

*The distilled notes per source. Deeper than Part 1; consult when executing a specific playbook chapter.*

---


# Google Search Central, Official Guidance (SEO Made Easy, Search Console Training, JavaScript SEO, Mythbusting, Ecommerce, Snippets)

> Source: Google Search Central YouTube playlists (official Google employees: Daniel Waisberg, Martin Splitt, John Mueller, Alan Kent, Aurora Morales). This is **authoritative guidance**, where it conflicts with practitioner advice, this wins. Transcripts: `D:\seo-courses\transcripts\google-*`.

---

## 1. How Search Works (mental model for builder agents)

Google operates in **four distinct stages**, treat each as a separate system to debug:

1. **Crawling**, Googlebot fetches URLs it discovered via links, sitemaps, and redirects.
2. **Rendering**, JS is executed in a second wave (queue can take days–weeks). Content only visible after JS execution arrives in the index late.
3. **Indexing**, content is parsed, deduplicated (canonical selection), and stored.
4. **Ranking/Serving**, queries are matched against the index.

**Debugging rule:** always use the **URL Inspection tool** in Search Console, not `site:` queries, not cache, not "view source". The live test shows the rendered HTML exactly as Googlebot sees it.

### Crawl rate, demand, budget, precise definitions

- **Crawl rate** = max concurrent connections Googlebot will use; auto-adjusts based on server responsiveness (slows on errors/slowness, `crawl-delay` in robots.txt is **ignored**).
- **Crawl demand** = how much Google *wants* to crawl (new URLs + estimated change frequency).
- **Crawl budget** = crawl rate × crawl demand. **Only matters for sites with ~millions of URLs or flaky servers.** For a typical site, "crawl budget problems" are actually *quality* problems, "crawled, currently not indexed" usually means Google crawled it and decided it wasn’t worth keeping.

### Crawl Stats report (Settings → Crawl Stats, domain properties only)

Watch: total requests, download size, **average response time** (HTML only, excludes resources/rendering), host status (robots.txt fetch failures = Google stops crawling entirely; DNS errors; server connectivity), breakdowns by response type / file type / crawl purpose (discovery vs refresh) / Googlebot type.

- A significant drop in crawl requests → check someone didn’t deploy a bad robots.txt, or the server is slowing down.
- Robots.txt must return 200 or 404. A 5xx on robots.txt fetch **halts crawling**.

---

## 2. Crawling & Indexing, rules that matter

### robots.txt
- Blocks **crawling**, not indexing. A robots-blocked page can still be indexed (and shown without a snippet) if other pages link to it. To keep a page out of the index: use `noindex` (requires the page to be crawlable) or HTTP auth.
- "Indexed, though blocked by robots.txt" appears as *Valid with warnings*, fix with noindex, not robots.

### Canonicalization (Mythbusting deep-dive with Martin Splitt)
- `rel=canonical` is a **hint/signal, not a directive and not a redirect**. Google scores duplicate candidates continuously (content fingerprinting, sitemap, site structure, internal links) and can flip its choice, if signals are ambiguous, the canonical can oscillate and Search Console data looks weird.
- **Do NOT use canonicals to group topically-related pages** or to "pool link equity" from unlike pages, it’s ignored and wastes crawl. Canonicalization exists to deduplicate *the same or near-same* content.
- Duplicates with slightly different content may still be deduplicated; sufficiently different pages ignore the canonical entirely.
- Google may serve a *different locale version* to users than the canonical (e.g., AT vs DE pages), expected behavior.
- For unavailable products: 301→similar product, or 404 (can come back). Don’t canonical a sold-out page to a different product.

### Index coverage report statuses
- **Error**, won’t appear in search (server errors, noindexed submitted URLs, 404s on submitted URLs). Sorted by severity × affected pages; fix top-down, then **Validate Fix**.
- **Valid with warnings**, indexed but something’s off (e.g., blocked by robots.txt).
- **Excluded**, not indexed, intentionally or correctly (noindex, duplicates, 404s, "crawled/discovered, currently not indexed" = quality/value signals).

### Old 404s (Mueller): 404s that have no traffic and no backlinks are **fine**, they don’t harm the rest of the site. GSC sorts crawl errors by importance; if the top ones are irrelevant, everything below is too. Check logs/analytics for traffic and GSC for links; reinstate a 301 only where it matters.

### URLs
- `.html` vs extensionless: **irrelevant** to Google. Never churn URLs for cosmetics, every URL change is a migration that takes time to settle. Choose durable URLs and keep them.
- Subdomain vs subfolder: **both fine**. Pick what you’ll keep long-term; subdomains must be verified separately in GSC and are learned as somewhat separate crawl units (may or may not share crawl budget).
- Multiple H1s: fine, HTML5 templates legitimately use several; semantic markup still encouraged.
- Sitemap limits: 50k URLs per file; use a **sitemap index file** beyond that, and split sitemaps by section for per-section index counts. Auto-generate sitemaps server-side.
- **`lastmod` is only trusted if it corresponds to real changes.** Auto-bumped lastmod on unchanged pages trains Google to ignore it. ETag/Last-Modified HTTP headers and structured-data dates also feed change estimation.

### Redirects & migrations
- 301/302 → permanent vs temporary signal. Keep redirect chains short; update internal links to point at the destination directly.
- **Migrate one thing at a time.** Don’t simultaneously change domain + host + URL structure + design + content. During migration: update sitemap progressively, keep both servers stable, verify redirects, don’t swap robots.txt rules at the same moment.
- GSC Change of Address tool exists for domain moves; verify both properties.

---

## 3. JavaScript SEO (Martin Splitt series, the most important technical corpus here)

### The two-wave indexing model
1. **Wave 1:** Googlebot indexes the raw HTML immediately.
2. **Wave 2:** the page enters a render queue; JS executes in headless Chromium (WRS); rendered content is indexed later (can take days–weeks).

**Builder rule:** check "View Page Source", if critical content/links/metadata aren’t in the initial HTML, they’ll be indexed late. Put what matters in the first payload (SSR, SSG, or hybrid rendering).

### Hard rules for JS-built sites
- **Links must be `<a href="real-url">`.** Never `div`/`span` + JS click handlers, never hash-based routing (`#/page`), fragments are ignored. Use the **History API** with real URLs for SPA transitions.
- Every route must return **HTTP 200 with content** when loaded directly in a fresh session (test in incognito, Googlebot has no service worker or shared state; each URL is visited independently).
- Don’t ship `noindex` in the initial markup of a page that needs the render wave, it can prevent wave 2.
- Per-page `<title>` and meta description must be set (via framework head management), generic shell titles across all routes is a classic SPA failure.
- Use semantic HTML (headings, sections, `<img alt>`), crawlers and assistive tech rely on the same semantics.
- Rendering fetches subresources against crawl budget: **cache JS/CSS/API aggressively; never make content-loading API calls via POST** (uncacheable → render failures at scale).
- SPA + no SSR? Options ranked: SSR/SSG > hybrid > **dynamic rendering** (serve bots a pre-rendered version, e.g., Rendertron middleware keyed on user-agent; cache rendered output; keep it warm). Dynamic rendering is a workaround, not a strategy, move to SSR when feasible.

### Testing toolkit
URL Inspection (live test + rendered HTML + screenshot), Rich Results Test (structured data eligibility), Mobile-Friendly Test, Lighthouse (lab data, predictive, not what users see), CrUX/PageSpeed Insights (field data, what counts).

---

## 4. Page Experience & Core Web Vitals

- Signals: **HTTPS, mobile-friendliness, no intrusive interstitials, Core Web Vitals** (LCP ≤2.5s, INP ≤200ms [replaced FID in 2024], CLS ≤0.1).
- Speed is a **real but small** ranking factor, it can break ties between near-identical pages (e.g., same supplier product descriptions) but won’t lift bad content. "Content still wins" (Splitt).
- **Lighthouse score is not a ranking factor**, it’s lab data from your machine. Ranking-relevant data is **field data** (CrUX, visible in PageSpeed Insights and GSC Core Web Vitals report). Test on the devices/networks your real users have (cheap Android on 3G ≠ your iPhone).
- ~53% of mobile sessions abandon at >3s load (Google data cited); average page then was ~15s, huge UX headroom.
- Fixes that matter: properly-sized compressed images (WebP/AVIF, explicit dimensions to prevent CLS), lazy-load below-fold (`loading="lazy"` now native), CDN + correct caching headers, less JS (JS must download+parse+execute, most expensive resource), enough server capacity.
- Lighthouse "you’ll save 6s" estimates are per-issue and overlapping, fixing one rarely yields the full estimate.

---

## 5. Search Console, the operational playbook

**Setup:** verify domain property (DNS record, covers all subdomains/protocols). No old account needed; just re-verify and add users with appropriate roles. Fix sitemap submissions; set up the **Insights report** for quick wins.

### Performance report, the daily driver
- **Metrics:** clicks, impressions, CTR, average position. **Dimensions:** queries, pages (canonical), countries, devices, search appearance, dates.
- Four core analyses:
  1. **High impressions / low CTR** → improve title links & snippets (this is the highest-ROI fix).
  2. **Expected queries missing** → content gap, build relevant content.
  3. **Important pages missing** → inspect URL for indexing issues.
  4. **Branded vs non-branded split** (regex filter on brand terms) → measures brand demand vs generic discovery.
- Filter by search type (Web/Image/Video/News), device, country; compare periods with same day-count/same weekdays.
- Caveats: anonymized queries exist; site-aggregated vs URL-aggregated metrics differ; queries table is capped → use API or bulk export for full data.

### URL Inspection
Three actions: **Test Live URL** (only place to see the live rendered page as Googlebot), **Request Indexing** (after meaningful changes/fixes), **View Crawled Page** (raw HTML + HTTP response + loaded resources). Indexed version ≠ live test result, check both.

### Traffic-drop diagnosis flowchart (episode 024)
Shape of the drop tells the cause:
- **Cliff to zero across all channels** → site-level technical (server down, bad deploy, robots.txt).
- **Gradual page-level decline** → page issues (rogue noindex), core-update re-evaluation, or lost relevance.
- **Recurring dips** → seasonality (compare to same period last year, not last week).
- **Spike-and-return** → reporting glitch (check chart annotations).
- Steps: extend to 16-month view → compare equivalent periods → segment by query/page/country/device/appearance/search-type → check Google Trends for demand shifts → check for new competitors cannibalizing brand terms.

### Other reports
- **Sitemaps report:** per-file index counts, split by section for granular coverage tracking.
- **Removals:** temporary (~6-month) hide of URLs; not a substitute for noindex/404.
- **Manual Actions:** Google "doesn’t hold a grudge", fix, submit reconsideration, action lifts. Site returns to its *natural* position, not the artificially-inflated pre-penalty one.
- **Security Issues:** hack/malware flags.
- **Crawl Stats:** see §1.
- **API + bulk export (BigQuery/Looker):** for data beyond UI limits (rows caps, 16-month window), export periodically to keep history.
- **Bubble chart analysis (ep 025):** plot queries as CTR vs position vs clicks, find "high impression, position 4–10, low CTR" quick wins.

---

## 6. Structured data & appearance

- Structured data makes pages **eligible** for rich results, never guaranteed. Validate with Rich Results Test; monitor in GSC enhancement reports; keep markup matching visible content.
- Breadcrumbs help Google understand site structure even without rich-result display.
- **Site names:** use WebSite structured data (`name`, `alternateName`) + clear site identity.
- Product markup on ecommerce pages: name, image, price, currency, availability, reviews, SKU, enables product rich results + Merchant Center feed integration (Search Console → Shopping tab listings).

---

## 7. Ecommerce SEO essentials (Alan Kent’s 8 tips, condensed)

1. **Technical basics first**, crawlability, clean URL structure (see "Designing a URL structure for ecommerce"), GSC verification. Platform stores: most handle this, but verify.
2. **Cover the whole shopping journey**, homepage (browsers), category pages (comparers), product pages (buyers), plus guides/reviews/comparisons for early-stage researchers. Write *own* product descriptions, supplier copy makes you a duplicate competing on speed/authority alone.
3. **Product variants:** each variant gets a unique URL (query param is fine); pick a canonical; canonical page must textually describe all variants (colors/sizes) so it matches variant queries.
4. **Recurring events (Black Friday, Mother’s Day):** **reuse the same URL every year**, never `/sale-2024` → `/sale-2025`. Keep the page live year-round with last/next offers; update content before each event. This accumulates link equity and index trust.
5. **Performance matters**, especially as tiebreaker vs identical supplier-description competitors.
6. **Patience**, SEO takes months; meanwhile diversify: email, social, reviews, offline QR codes.
7. **Paid link schemes = guideline violation**, negative repercussions possible.
8. **Optimize for users, not algorithms**, algorithms change, the goal doesn’t. Bounce/engagement signals tell you if you’re attracting the wrong audience.

Merchant Center: onboard product feeds via Search Console shopping tab; free listings surface products in the Shopping tab.

---

## 8. Snippets & metadata (SEO Made Easy)

- **Title links:** unique, descriptive, concise; Google may rewrite from H1/content if titles are poor, write them well and rewriting is rare.
- **Meta descriptions:** Google often generates snippets from page content matching the query, but good meta descriptions still win clicks for stable queries, one unique description per page, ~160 chars, answering "why this page".
- **Image optimization:** descriptive filenames + alt text; correct display dimensions; compress + modern formats; images rank in Image search and feed understanding.
- **Localization:** hreflang requires valid language codes, **reciprocal pairs**, self-referencing entries, and an x-default; point internal links at the user’s language version.
- **HTTPS:** ranking signal + user trust baseline.
- **Internal links:** descriptive anchors (never "click here"); link to canonical URLs directly.

---

## 9. Official myths, busted

| Myth | Reality |
|---|---|
| Crawl budget affects every site | Only ~millions-of-URLs sites or flaky servers |
| `crawl-delay` slows Googlebot | Ignored, Google auto-adjusts to server health |
| Canonical = redirect | It’s a hint for *duplicate* pages only |
| Canonical groups topics/pools PageRank | No, deduplication only |
| "Crawled not indexed" = crawl problem | Usually a **quality** problem |
| Lighthouse score = ranking | Lab tool; field data (CrUX) is what matters |
| AMP is required/ranking factor | Never was; speed matters, AMP doesn’t |
| Manual actions blacklist forever | Lifted on fix; no grudge, but position may differ |
| Old 404s hurt the site | Harmless if no traffic/links |
| `site:` count is accurate | Rough sample, use GSC index/sitemap counts |
| Subdomains rank worse / subfolders always better | Both fine, pick for durability |
| One H1 per page required | Multiple H1s fine (HTML5) |
| .html URLs rank worse | Extensions irrelevant |

---

## 10. Builder-agent checklist derived from official guidance

- [ ] Verify GSC domain property; submit sectioned XML sitemaps (+index file if >50k).
- [ ] Critical content + links + titles + meta in initial HTML (SSR/SSG), or configure dynamic rendering as interim.
- [ ] All navigation via `<a href>` real URLs; History API for SPA transitions; no hash routing.
- [ ] Direct-load test every route in incognito → expect 200 + full content.
- [ ] Canonicals only on true duplicates; internal links point at canonical URLs.
- [ ] `noindex` for exclusion (not robots.txt); robots.txt returns 200/404.
- [ ] Real `lastmod` in sitemaps; ETag/Last-Modified headers correct.
- [ ] Image: sized, compressed, WebP/AVIF, alt text, lazy-load below fold, explicit dimensions (CLS).
- [ ] HTTPS everywhere; mobile-friendly; no intrusive interstitials.
- [ ] Structured data validated in Rich Results Test for eligible types (Article, Product, Breadcrumb, FAQ, Organization, WebSite/site-name).
- [ ] Recurring promo pages: stable evergreen URLs updated in place.
- [ ] Product variants: unique URLs + canonical + canonical page describes all variants.
- [ ] Monthly: Performance report (CTR wins, missing queries), Index coverage errors, Crawl Stats anomalies, Core Web Vitals field data.
- [ ] Quarterly: bubble-chart CTR analysis; 16-month trend; content-refresh candidates.
- [ ] Escalate: unexplained coverage drops, manual actions, persistent 5xx spikes, canonical flapping.


---


# COURSE: Ahrefs SEO Course for Beginners (Sam Oh), 14 lessons, ~2h
Source: https://www.youtube.com/playlist?list=PLvJ_dXFSpd2vk6rQ4Rta5MhDIRmakFbp6
(also as single video: https://www.youtube.com/watch?v=xsVTqzratPs)
Transcripts: D:\seo-courses\transcripts\ahrefs-seo-beginners\

## L1, SEO Basics ([video](https://youtu.be/btwC4zmewss))
- SEO = optimizing content to be discovered via organic search. Why SEO wins: free traffic, consistent (vs social/email spikes), massive reach (~4B Google users).
- How Google works: (1) crawlers ("spiders") start from seed URLs, follow links, feed the search index; (2) ranking algorithm with hundreds of signals, ~500-600 algo tweaks/yr.
- Top 3 foundational ranking factors:
  1. **Backlinks** = votes; Google: "if other prominent websites link to a page, that’s a good sign the information is well trusted." Correlation between unique-domain backlinks and organic traffic is proven.
  2. **Search intent** = reason behind query; read it from top-ranking pages (e.g. "slow cooker recipes" → blog listicles; "slow cooker" → ecommerce category pages). Don’t fight the SERP.
  3. **Content depth** = cover what searchers need ("how to drive a car" → seatbelt, pedals, mirrors). Depth ≠ length, "turn off iphone 12" top page = 185 words and wins.

## L2, How to Choose Keywords: 5-point checklist ([video](https://youtu.be/H-B5oFJjL8I))
1. **Search demand**, search volume exists (but can mislead; e.g. "km to miles" = 478k/mo but 80% of searches end without a click).
2. **Traffic potential**, total traffic the #1 page gets incl. all its keywords. Avg top page ranks for ~1,000 keywords in top 10 (Ahrefs 3M-kw study). Judge by top page’s traffic, not the kw’s volume.
3. **Business potential**, score 1-3: 3 = ready to buy ("buy used golf clubs"), 2 = pre-purchase research where you can plug product ("best golf clubs"), 1 = informational only, 0 = skip.
4. **Match search intent** (below).
5. **Can you actually rank?** (below).

## L3, Search intent = the 3 C’s ([video](https://youtu.be/zXPZ0L2cOw4))
- **Content type**: blog post / product / category / landing page / video, match the dominant type in SERP.
- **Content format**: listicle, how-to, step-by-step, opinion, tool/calculator.
- **Content angle**: the hook/benefit, e.g. "freshness" (current year in titles) for "best X" queries. Least important but still a signal.
- If top pages are category pages and you have a blog → can’t match intent → skip the keyword.

## L4, Generating keyword ideas ([video](https://youtu.be/GsW5GeDXNkU))
- Process: (1) generate ideas → (2) validate. Tools: Ahrefs Keywords Explorer / free Ahrefs Keyword Generator.
- Seed keywords = broad niche terms → Phrase match report → thousands of ideas.
- Filters: volume ≥300 → check SERP traffic potential → business potential score → intent match.
- **Modifiers shortcut**: filter include-list of modifiers to surface intent-matching keywords. Commercial modifiers: best, top, vs, review. Informational: how, what, who, where, why, guide, tutorial. Use broader seeds when adding modifiers.
- **Competitor-driven ideation**: keyword list → Traffic share by domains report → organic search competitors (not business competitors) → their Top Pages report → steal topics that don’t contain your seeds → also mine new seeds from it (repeat loop).

## L5, Ranking difficulty ([video](https://youtu.be/TPtCjy4n4cU))
- Competitors = pages ranking for THAT keyword (not business rivals). 3 checks:
  1. **Intent weakness**: do some top pages fail to match intent? Check if titles/URLs contain the kw or close variant; mixed-intent SERPs (e.g. "best convertible car seat for small cars" with generic pages) = opportunity.
  2. **Metrics**: referring domains to top pages (can I get more quality links?); Domain Rating, attack keywords where your DR is in the same ballpark as at least some top-10 sites (low-DR sites in top 10 = good sign).
  3. **Topical authority**: niche sites outrank big-DR generalists on their topic (plumbing site DR42 outranking giants for "unclog toilet"). Ask: is my site equally/more topically authoritative? Judge by domain name + About page.
- Rule of thumb: more "yes" answers to the 3 questions = higher odds.

## L6, What on-page SEO is NOT ([video](https://youtu.be/lu9-GvQGZAE))
- NOT exact-match keyword stuffing (Google understands connectors/synonyms; Healthline #1 for "diet plan" w/o the phrase).
- NOT keyword density (top page ranks ~1,000 related keywords, can’t mention all 3x).
- NOT a minimum word count (628-word backlink checker page gets 130k visits/mo; 76-word image page gets 170k/mo).
- IS: satisfying searcher intent + tangible items (titles, subheadings, internal links, readability, images).

## L7, Creating a page optimized for search ([video](https://youtu.be/IrFAeQgzE7w))
1. Open top 3 relevant (intent-matching) pages. Find shared subtopics, format, angle, audience assumptions (e.g. all "best golf club sets" posts target beginners, mention Callaway Strata + Wilson Staff, organize by categories, subheads = product names).
2. **Page-level content gap**: Ahrefs Content Gap → enter 3 competitor URLs → "2-3 intersections" → keywords ≥2 of them rank for (1 in top 10) = subtopics & audience language you must cover.
3. Technical on-page checklist:
   - Target keyword (or natural variant) in title.
   - Short descriptive URL slug (usually the keyword, hyphens).
   - Meta description, not a ranking signal but affects CTR; Google rewrites ~63% of the time anyway (192k-page study) so don’t over-invest.
   - **Internal links** to/from relevant pages (pass authority, help Google understand). Find opportunities: `site:yourdomain.com <topic>` or Site Audit link explorer.
   - Images: descriptive filename, useful alt text (keywords in context, no stuffing), compress (ShortPixel), speed is a ranking factor.
   - Readability: short sentences/paragraphs, descriptive subheads, large font, no big words, write like you speak. Hemingway app → ≤ grade 6.
   - OG tags + schema markup (recipe/faq/etc for rich results; RankMath/Yoast plugins).

## L8, Link building intro ([video](https://youtu.be/oFyF417Q0oM))
- Link building redefined: "building relationships with relevant site owners who want to link to your content because it enhances theirs", relationships, relevance, value exchange.
- Backlinks = prominent Google ranking signal since PageRank (1998); clear correlation with traffic in 1B+ page study. Needed for competitive (high-traffic/revenue) phrases.

## L9, 3 ways to get backlinks ([video](https://youtu.be/5lkzCLsOKrI))
- **Create** (directories, comments, profiles): easy, mostly ineffective.
- **Buy**: against Google guidelines, penalty risk, ~$353/link avg (250-site study). Not worth it long-term.
- **Earn**: (a) email outreach, most common; (b) become a cited source for journalists; (c) organic/passive, unreliable until famous.
- Rule: the harder a link is to get, the more valuable it is.

## L10, 5 attributes of a quality backlink ([video](https://youtu.be/Yf1Bc3JsTSA))
1. **Relevance**, page-level AND domain-level topical fit; ideal link sits inside body content as a genuine recommendation.
2. **Authoritativeness**, page’s link power (PageRank: more in, fewer out = more passed). Ahrefs proxies: DR (domain), UR (page).
3. **Anchor text**, helps Google understand topic; keyword-rich anchors at scale = link scheme penalty. Natural earned anchors are mostly brand/URL/"click here", over-optimization rarely a worry when earning.
4. **Rel attribute**, nofollow/UGC/sponsored are now "hints" (may pass value); followed links proven to pass authority → aim for followed.
5. **Placement**, prominent in-content links > footer/sidebar links.
- Ideal link: relevant + authoritative + followed + descriptive anchor + in-content.

## L11, 3 beginner tactics + process ([video](https://youtu.be/jTABfFhvFpE))
- Process = **prospecting** (find many relevant candidates) → **vetting** (check quality/relevance) → **outreach** (email).
- **HARO/journalist sourcing**: sign up as source → filter emails by keyword (Gmail filter on haro@helpareporter.com) → respond fast, short, lead with expertise, follow their instructions exactly. Value exchange: expertise ↔ mention+link. (Note: HARO now operates as Connectively/other journalist platforms, same workflow applies.)
- **Guest posting**: find sites via `intitle:"write for us" + <niche>` in Google, or Content Explorer (topic + DR filter 30-60 / 10-30 if new, one page per domain, sort by author count = likely accepts guests). Vet for non-spammy + stable organic traffic (declining traffic = possible penalty, skip). Pitch a specific missing topic with proof (e.g. site has "best balls for kids/distance" but not "for seniors"). Target ~10 prospects per post you can write weekly.
- **Skyscraper**: find heavily-linked content → make something better → pitch its linkers.

## L12, Blogger outreach that works ([video](https://youtu.be/GK049ajJWRQ))
- Objective: convince people with audiences to talk about you. **Sniper > shotgun**, never blast templates.
- Who to email: author if they control the site’s content, else editor (About/Team page, "write for us", LinkedIn). Find emails: Contact/About pages, Hunter.io name+domain.
- **Anatomy of a converting outreach email**:
  1. Subject: accurate + curiosity ("New data: best balls for high-handicappers").
  2. Intro: why you’re emailing, get them to read on ("noticed you cover X and Y but not Z").
  3. Qualification/justification: why YOU (spent $ testing balls on launch monitors).
  4. Pitch: the ask + value proposition (free data → helps their sales). Skyscraper example: pointing out a stale stat on their page + offering fresher one, helped them keep content current → 27 links without even asking for a link.
  5. End with a conversation opener ("Is that something you’d be open to?"), not a cold close.
- Only pitch your BEST content; mass outreach on mediocre content burns bridges.

## L13, Technical SEO essentials ([video](https://youtu.be/_bewU9AvKwU))
- **noindex**: commonly left behind after redesign migrations → pages vanish from index. Check after every redesign. Deliberately noindex low-value pages (author archives etc.).
- **robots.txt**: per-(sub)domain rules; User-agent + Disallow. `Disallow: /` for Googlebot = dead site. Common legit use: block /wp-admin/.
- **XML sitemap**: lists important URLs; auto-generated by Yoast/RankMath; reference via robots.txt Sitemap directive + submit in GSC.
- **Redirects**: consolidate signals from old/duplicate URLs (redirect old "best-golf-balls-2018" → current).
- **Canonical tag**: preferred-URL hint for duplicates (http/https, params); Google may ignore it (e.g. prefers HTTPS anyway).

## L14, Technical best practices / maintenance ([video](https://youtu.be/RFlpwKQ0bEs))
- **Site structure**: logical hierarchy (home → main topics → subpages) via internal links, helps crawling + topical understanding.
- **Speed**: only "slowest experience" pages get hit, but cache content + compress images as baseline.
- **Ongoing maintenance**: ~50 common issues (broken internal links, orphan pages, duplicates, redirect chains), run scheduled site audits (Ahrefs Webmaster Tools free: 10k pages/site, weekly crawls, flags issues + fixes).


---


# COURSE: Ahrefs Technical SEO + SEO Audit tutorials (10 vids, Sam Oh / Joshua Hardwick / David McSweeney)
Transcripts: D:\seo-courses\transcripts\ahrefs-technical-seo\

## Indexing foundations (every page must pass these before anything else matters)
- **noindex tag**, removes/prevents pages in Google’s index. Legit uses: author archives, thank-you pages, thin utility pages. Check after every redesign, devs often leave `noindex` on.
- **robots.txt**, at root domain; per-subdomain files allowed. Directives: `User-agent` (crawler name; `*` = all) + `Disallow` (path). `Disallow: /` = de facto death. Use for /wp-admin/ etc. First place to check for indexing problems.
- **XML sitemaps**, auto-generate via Yoast/RankMath; add `Sitemap:` directive in robots.txt AND submit in GSC.
- **Redirects (301)**, consolidate duplicate/old pages (e.g. /best-golf-balls-2018 → /best-golf-balls) to pass link signals.
- **Canonical tag**, `rel="canonical"` declares preferred URL version; consolidates signals, but Google CAN ignore it if contradictory (e.g. canonical→HTTP while HTTPS exists).
- **HTTPS**, ranking signal since 2014, part of page-experience signals. Free TLS via Let’s Encrypt (DV cert fine for most); OV/EV for stricter identity. Migrate fully: every http/https/www/non-www permutation must 301 to ONE canonical version, sitewide, not just homepage.

## Core Web Vitals (now part of Page Experience signals)
Metrics assessed at **75th percentile of real users**, mobile & desktop scored separately (mobile-first indexing = mobile scores are what count).
- **Field data** (CrUX, real Chrome users, rolling 28-day average, changes take ~4 weeks to show) vs **lab data** (Lighthouse/DevTools, instant but synthetic). Iterate on lab, confirm in field.
- **LCP** (largest visible element, usually hero image/H1): target <2.5s. Top causes: slow server response, render-blocking JS/CSS, slow resources, client-side rendering. Fixes: preload hero image (`<link rel="preload">`), good hosting, caching, CDN, image compression+lazy-load below fold.
- **CLS** (visual stability, worst 5-second window): target <0.1. Causes: images/iframes/ads without dimensions, injected content, late-loading fonts. Fixes: width+height attrs or CSS aspect-ratio boxes, reserve space for ads/embeds. Debug: webpagetest.org filmstrip → "Highlight layout shifts".
- **Interactivity**: FID was the original metric, **superseded by INP (March 2024)**, target <200ms. Causes: heavy JS blocking main thread. Fixes: defer/async attributes (`async` for early-needed scripts, `defer` for the rest), minify, remove unused JS.
- CSS/JS hygiene: minify, inline critical above-the-fold CSS, remove unused CSS (DevTools → Cmd menu → "Show Coverage" → red = unused), split non-critical to separate files loaded only where needed.
- Baseline quick wins before any of this: good host + caching + image compression + CDN.
- Reality check (per Google): the Speed Update "only affects the slowest pages", don’t obsess over milliseconds, just don’t be slow.

## Google-only audit (free, no tools, Glen Allsopp method)
Test every URL permutation redirects to canonical → then run `site:` queries:
- `site:domain.com`, sanity-check indexed count vs. real page count (finish crawl first if unsure)
- `site:domain.com inurl:www` / `-inurl:https`, stray protocol/subdomain versions in index (found a hacked site this way: HTTP pages still indexed + spammy injected links)
- `site:domain.com intitle:oldbrand.com`, template/title issues after rebrands/migrations
- `site:olddomain.com`, pages missed by migration redirects
- `site:domain.com inurl:page / inurl:sort / inurl:tag / inurl:author`, pagination, faceted-nav, WP archive bloat
- `site:domain.com inurl:welcome-to-wordpress OR inurl:hello-world` + `"lorem ipsum"`, default install / dummy content still indexed
- `site:domain.com filetype:pdf`, accidentally public docs (salaries, addresses found in real audits)
- Agency use: these queries double as prospecting (find businesses with obvious technical issues).

## The 16-step SEO audit (Joshua Hardwick, ~20 min)
Tools: GA, GSC, PageSpeed Insights, Structured Data test, Ahrefs (or Screaming Frog), Copyscape, SERPsim, word counter.
1. Canonical redirects, all 4 permutations 301 to one version.
2. Crawl, Site Audit w/ "Execute JavaScript" + "check external links" on.
3. `site:` count vs. crawlable pages, investigate discrepancies (e.g. Google indexing page fragments).
4. Brand-name SERP, rank #1? If not, big problem.
5. On-page spot-checks on top-traffic pages (GA landing pages or Ahrefs Top Pages): title length via SERP simulator + keyword-optimized (verify target KW has volume in Keywords Explorer), meta description length/quality, single H1 containing keywords.
6. Site Audit → HTML Tags report, multiple H1s, missing metas, long titles sitewide.
7. Duplicates, internal via Content Quality report; cross-domain via Copyscape (footer disclaimers trigger false positives, check context).
8. Thin content, low-word-count report.
9. Speed, Performance report "slow pages", sort by organic traffic to prioritize.
10. Structured data, Rich Results/validator per template type.
11. Organic traffic trend, GA acquisition 12-month view + landing pages.
12. Rankings trend, Organic Keywords graph overall + history per target keyword.
13. **Low-hanging fruit**, filter Organic Keywords: positions 3-5 (or 4-10) AND volume ≥1,000 → biggest ROI updates.
14. Backlink sanity, referring-domains graph trending up; anchors mostly branded = natural; sort RDs by lowest DR for spam sniff-test.
15. **Broken pages with links**, Best-by-Links filtered to 404 → reclaim via 301 or outreach. Also outgoing broken links report.
16. **Content gap**, Ahrefs Content Gap: 2-3 competitors rank, you don’t → filtered by volume = new content roadmap. Then **content audit**: prune/redirect low-quality zero-traffic pages, Ahrefs deleted 200+ posts and traffic went UP. (Caveat: only delete when certain it’s valueless; improve+relaunch is usually safer.)

## Audit workflow tips (Sam Oh, Problogger crawl)
- Health Score = % of crawled URLs without critical errors.
- Prioritize 404s by adding "dofollow backlinks" column → sort desc = reclaim the most link equity first. Export CSV → assign to team/freelancer.
- Trailing-slash duplicates (page vs page/) both live with no canonical = "bad duplicate"; fix via canonical tag AND correcting the internal link. Check "Number of Inlinks" to see which URL is the real one.
- Order of work: Errors → Warnings → Notices. Each issue in Site Audit includes a how-to-fix snippet.
- Schedule weekly crawls to catch new issues (broken pages, orphan pages, redirect chains, dupes) automatically.
- Penguin/Barracuda tool overlays Google update dates on traffic → fast penalty vs. seasonality check (watch for confounders like Christmas).


---


# COURSE: Ahrefs Keyword Research playlist (18 tutorials, Sam Oh + Kathryn Aragon)
Transcripts: D:\seo-courses\transcripts\ahrefs-keyword-research\

## Search intent, the dominant thing to get right
- 4 buckets: informational / navigational / commercial investigation / transactional. Navigational queries for other brands = skip (Bank of America example: pos-3 gets ~0 traffic). Exception: navigational queries where intent doesn’t require the brand itself (Wise ranks for "bank routing number" pages → 100k+ visits).
- **Fractured/mixed intent**, Google QRG calls them "queries with multiple meanings"; SERPs blend dominant/common/minor interpretations. Analyze: organic results (location-matched, incognito), SERP features, and **SERP position history**, stable top ranks = clear dominant intent & harder but predictable; volatile = Google unsatisfied, faster entry but uncertain.
- Ahrefs team tested intent interpretation, 3 experts disagreed on buckets but agreed on dominant intent. **Don’t argue buckets; serve the dominant intent via the 3 Cs**: Content type (blog vs product vs category), Content format (listicle/how-to/review), Content angle (freshness/"for beginners"/cheap). If you can’t match dominant intent → target common intent or walk away.

## The 3 keyword types every site needs (new-site workflow, file 006)
1. **Money terms**, core business drivers, hard to rank, nobody links to them.
2. **Linkable topics**, created to attract links → funnel PageRank via internal links to money pages.
3. **Low-competition**, rank without links → early traffic + natural link growth.
Workflow: brainstorm 5-10 seed topics → Keywords Explorer → build "relevant words" list from autocomplete/related searches/Google Correlate → Include-filter (Any word) the giant list → segment into the 3 buckets via modifier lists (informational: how/what/tutorial/ideas; commercial: best/top/review/vs; transactional: buy/hire) → volume tier filters (≥1000, then 500-999…) → check intent on SERP button → assess difficulty.

## Keyword Difficulty, read it right (files 003, 005, 015)
- Ahrefs KD = weighted avg of referring domains to top-10 pages. Links ONLY, no on-page, relevance, brand, or authority factors. Scores differ across tools; no metric can decrypt 200+ signals.
- **False positives**: "compact flash card reader" KD-low but Amazon/BestBuy/B&H dominate = **brand equity** wall (transactional head terms). "media storage" KD 3, same story. Skip unless you’re in their league.
- KD is **relative to YOUR site**: DR95 site finds KD5 easy; DR15 doesn’t. Judge top-10’s DR spread, look for a weak site already ranking.
- Quality > quantity: SERP may show a 4-RD page outranking a 209-RD page → inspect dofollow links sorted by ascending DR for spam; WikiHow example: first 70-80 RDs were spammy blogspot vs competitor’s clean editorial links. Relevance of linking sites matters.
- **Use KD both directions**: low-KD+questions report = rankable info topics; HIGH-KD topics = concentrated link prospects (target them to harvest links → funnel PR internally, not necessarily to rank).
- Always eyeball the SERP before committing. YMYL (health/wealth) → weigh E-E-A-T harder.

## Traffic potential beats search volume (files 009, 013, 010)
- Ahrefs study (3M queries): avg #1 page ranks for ~1,000 more top-10 keywords; 64% of #1 pages rank for multiple 1K+-volume keywords; pages with stronger backlink profiles (UR) rank for more keywords; longer content ranks for more keywords (but write for intent + next questions, not word count).
- → Judge topics by the **traffic of the current top page** (Top Pages report / SERP overview traffic column), not the keyword’s volume. "bumps on skin" vol 9.6k → top page gets 90k visits/mo. Broad head terms often get LESS traffic than volume suggests.
- **Parent topic**: rank for the general topic, inherit the long-tail. "how to bbq steak" (350/mo) → parent "how to grill steak" (11k/mo, Canada-heavy!).
- **Clicks & clicks-per-search** (Ahrefs-only metric): "Donald Trump age", high volume, near-zero clicks (knowledge card kills them). "iPhone 7 case" CPS >1 = multi-click comparison behavior = rank #4 still pays. Check paid-vs-organic click % and CPC as commercial-intent proxy. Return Rate = repeat-search behavior hint.

## Finding untapped/low-competition keywords (files 002, 012)
- **Competitor Top Pages + KD≤20 filter** → verify slug matches top keyword → check caret for true global traffic → chain to more competitors via Competing Domains.
- **Modifier expansion**: find geo/audience modifiers that CHANGE the SERP ("first time home buyer" + 50 states → paste comma-separated list into Include/Any-word filter → 50 related low-comp pages, good site structure too).
- **Uncommon seeds**: competitor’s Organic Keywords filtered "doesn’t contain [obvious seeds]" → discover words like "stubble", "goatee", "mutton chops" → feed THOSE as new seeds.
- **vs/review harvesting**: scrape ecommerce brand facets (Scraper extension → "Scrape similar") → paste brands into Keywords Explorer → Include "vs, review*" → near-zero-competition affiliate keywords.
- **Topical long-tail** (file 012): topics that aren’t part of a bigger topic ("keyword cannibalization", "black shoes with navy suit"), tiny volumes but near-zero competition; 20-50 such pages = cumulative traffic. Find via modifier-within-topic Include filters ("with"), location suffixes ("rent a truck in"), question reports. Litmus: top page gets way more traffic than top keyword’s volume + low RDs.
- **Top Pages method for long-tail topics** (012): big competitor → Top Pages → filter top-keyword positions 1-3 + volume ≥1k → pages ranking for many 1k+ keywords = deep topics worth covering.

## Google Keyword Planner & Trends hacks (016, 017)
- GKP: exact-ish volumes via **impressions** (set max CPC bid → plan overview shows impression counts); paste a competitor URL or whole domain for keyword ideas (not their actual rankings, topical suggestions only); exclude brand via keyword-text filter; question filters (who/what/how...); **sort by top-of-page bid high-range** = commercial intent map; set location to your city/service area for real local volumes; regional breakdown → market research for expansion/hreflang decisions.
- Google Trends: 5-year view to separate fad vs seasonality vs growth (selfie stick = declining seasonal fad); compare brand-vs-brand by region for global SEO targeting; YouTube-search filter for video topic timing; rising "breakout" queries for early-mover topics. Pair with a real KW tool, Trends shows direction, not volume.
- Free-tool stack reality check (018): Keywords Everywhere (GKP bucketed data, singular/plural show same fake volume), niche autosuggest (Etsy/Amazon/Pinterest), Answer the Public (~170 questions vs Ahrefs' 93k). Fine for ideation; insufficient for intent/clicks/difficulty decisions.


---


# COURSE: Ahrefs Advanced Link Building Course (Sam Oh), 15 lessons
Source: https://www.youtube.com/playlist?list=PLvJ_dXFSpd2t31nRDgkPZ7p1PDfHmaW8P
(full single video: https://youtu.be/fNUxzKZE6K8) | Transcripts: D:\seo-courses\transcripts\ahrefs-advanced-link-building\

Core framework: **Seed → Lookalike → Segment** prospecting, then vet → blitz list → hybrid outreach.

## M1, Fundamentals
**L1.1 Role of content in link building**, Content dictates pitch angles; informational content (how-tos, data studies) is far easier to get links to than commercial/product/affiliate pages. Why people link: (1) reference/support a point, stats are #1; (2) delegate detail ("read more here"); (3) ego/self-promotion (ego bait: mention them positively, Inc. 5000 gets 1000s of links/yr); (4) implicit/explicit relationships (superfans + friends link to each other). Universal drivers: prove you’re not making stuff up, add credibility, trust.

**L1.2 Competitor backlink analysis**, Find competing pages via top-10 SERP for target query, OR for linkable assets use Content Explorer (topic + ≥100 ref domains + exclude subdomains + <1000 page traffic to avoid obvious SEO targets). To find *linkable points*: (1) Anchors report, repeated anchor = resonating point (ignore branded/URL anchors); watch for redirect-inherited links (Point Blank SEO → 130 RDs = irrelevant) and promises the page doesn’t keep ("Download ebook" anchor but no ebook = angle); (2) referring page *titles* for commonality (e.g. all "free SEO ebooks" listicles); (3) page content for incorrect/outdated claims (LSI keywords page: 500+ RDs linking to wrong info → pitch correction). Follow footprints from seed → bigger prospect pool (intitle:"best" intitle:"seo" intitle:"ebook" / Content Explorer title search; widen seed term: "seo"→"marketing").

## M2, Prospecting
**L2.1 Seed prospects** = people who link to a page for a specific reason (the "linkable point"). The point reveals both prospects AND pitch angle.
**L2.2 Lookalike prospects**, like ad-platform lookalike audiences: find pages sharing the seed’s footprint. Tools: Ahrefs (Site Explorer anchors/surrounding-text search, Content Explorer title/phrase search, Batch Analysis), Google intitle: queries + SEO toolbar, Scraper Chrome extension ("Scrape similar" → XPath). Real examples: 1,800 prospects via "%" anchor footprint; YouTube embeds/course-video linkers as warm prospects (unlinked-mention style conversion).

**L2.3 Segmentation**, Segment prospects BY LINKABLE POINT (not by metrics). Each segment = one tailored email that stays personalized at scale (all members share the same reason for contact). E.g. "mentioned 93% stat" segment → identical email works for all 800. Automate ongoing prospecting: Ahrefs Mentions/Backlinks Alerts on footprints (e.g. alert on "93% of online experiences" phrase → new prospects emailed to you).

## M3, Vetting
**L3.1 Three-level vetting**: (1) metrics filter, DR (imperfect, page-level PR anyway) + domain traffic (DR90 w/ 0 traffic = likely penalized, skip); (2) page relevance, scan titles, keep borderline-relevant domains, drop unrelated languages/topics; (3) **pitch-angle validation**, confirm the page still mentions the linkable point. Scale it with Screaming Frog custom search (List mode + "Contains 93" in Page Text → 779/875 matched in minutes vs 8h manual). Complex/contextual points need manual vetting.
**L3.2 Who to contact**, author, editor, marketing/content manager, or webmaster; generic@ last resort. Match to goal: guest post → editor/head of content (not author); edit existing post → author or editor. Measured link-acquisition rates: authors 7.56%, editors 7.55%, generic 4% (small co’s), other titles 1.64%. → Contact the person who can actually add the link.

**L3.3 Blitz list**, fully ready-to-send sample of vetted prospects. Purpose: skip slow email-finding AND validate the campaign before full investment (know your conversion rate in <1h). Method: Hunter Author API (or Hunter for Sheets add-on) on prospect URLs → Neverbounce API/Clean to validate deliverability → ~60-120 valid emails per 1000 prospects in 15-30 min. Implementation options: Sheets+AppScript, Hunter-for-Sheets+Neverbounce-CSV+VLOOKUP, custom code (~$50 dev), or Sheets+Zapier. Don’t farm email-finding to random freelancers (bad quality).

## M4, Outreach
**L4.1 Three approaches**: Shotgun (same email to all, fast but burns bridges/domains/brand), Sniper (bespoke emails, best links/relationships but unscalable, months per link), **Hybrid (recommended)**, segmented personalized templates; typical conversion **5-12%**.
**L4.2 Value props**, linkable point = reason for contact; brainstorm value by asking "how does this impact the prospect and THEIR audience?" Strongest example: pages recommending eating cinnamon for weight loss (dangerous, cinnamaldehyde) → correction pitch hit ~20% conversion at big publications. SEO-stats "outdated 93% stat" pitch = 6.8% (no follow-ups); improvement idea = offer to rewrite their paragraph whether or not they link.
**L4.3 Writing non-templated templates**, Write ONE real personalized email first, then extract merge fields, never the reverse. Exercise: draft like writing to a friend → edit out empty statements/false flattery/criticism-framed-as-question → get expert feedback. Bibi Raven’s notes: center the email on THEIR audience not you; avoid vague "etc."; pitch with enthusiasm + results others got; confident not conditional ("if interested" → weak); vary sentence openers; warm creative sign-off. Templatize by segment → only {name} needs to change.

## M5, Team, systems, workflows
**L5.1 Team roles**, specialized assembly-line beats all-in-one marketers: **Prospector** (SEO-strong, analytical, creative; works with content/design/dev; often doubles as PM), **Vetter** (detail + consistent instruction-following; bulk of labor; best hire came from SaaS QA, no SEO exp needed), **Outreach manager** (communication, people skills, copywriting; sales background works). Hire for core competencies, teach the SEO.
**L5.2 System components**: (1) specialized team structure; (2) shared specific goal (e.g. "50 new RDs/mo to revenue pages") + team bonuses; (3) SOPs, document the process yourself as if teaching a total beginner, video+docs; (4) PM tool w/ Kanban (Brainstorm→Prospect→Vet→Outreach→Done); (5) Google Sheets as shared campaign DB (QUERY/FILTER/VLOOKUP + AppScripts + API integrations; dedupe by root domain via Data→Data cleanup; pre-filter spam footprints /job/ /forum/ .ru .pw blogspot; drop <200-visits domains).
**L5.3 Triangular workflow** (beats linear): prospector seeds+segments sheet → hands to outreach mgr (drafts segment templates in parallel) + vetter (validates blitz batch first) → blitz emails go out FAST to validate angle → if conversion good (e.g. 9.5%), vetter finishes full list manually, feeding outreach manager daily batches. Prospector moves to next campaign immediately. Team of 3 → hundreds-thousands of personalized emails/week. Never use the machine to spam, vet extra, personalize extra.


---


# COURSES: Ahrefs Local SEO (3 vids) + Ahrefs YouTube SEO (8 vids)
Transcripts: D:\seo-courses\transcripts\ahrefs-local-seo\, ahrefs-youtube-seo\

## LOCAL SEO
**Why:** 46% of Google searches are local; 88% of mobile local searchers visit a store within a week; Snack Pack gets ~33% of clicks, local organic ~40%. **Rank in both.**

**Google Business Profile (step 0-1):** Mobile-friendly site first (61% more likely to be contacted). Claim at google.com/business → exact NAP. Rules: no keyword-stuffing the business name ("Coffee Shop Downtown 24/7" = violation); service-area businesses can hide home address; pin marker exactly; primary category = "my business IS a ___" not "has ___"; add secondary categories, hours, deep links, real photos, description. Verify by phone/mail.

**Local keyword research:** list all services → expand via competitor service menus + Keywords Explorer Matching Terms → **verify local intent** (search each: map pack + local organic present? Bulk: SERP-features filter "local pack") → national-level volume for prioritization (city-level data is unreliable; GKP location settings can approximate) → **group into pages by SERP overlap**: "oil boiler installation" SERP = all dedicated pages → needs own page; "gas boiler installation" SERP = generic boiler pages → covered by parent page; no map pack + all DIY results = no local intent, skip as landing page.
- SiL = "Service in Location" keywords. Idea sources: autocomplete, Craigslist services section, and **competitor organic keywords** (exclude their brand → reveals things like "coffee shop near Eaton Centre").
- Service pages: NO keyword stuffing. Mine "things to mention" from (a) top-ranking competitors' talking points (symptoms, procedure length, costs, options), (b) People-Also-Ask/Buyer’s-guide questions, (c) competitors' organic keywords top-10 (reveals neighborhoods/zip codes to mention).

**On-page for local:** single-location → optimize HOMEPAGE for "service + city" (NAP in footer sitewide, must match GBP exactly; LocalBusiness schema via Structured Data Markup Helper; testimonials). Multi-location → location directory → city pages → branch pages w/ unique NAP + hours + local keywords (see Enterprise’s URL hierarchy). Standard on-page (KW in title/H1/slug/short URLs) still applies.

**Local link building**, aim for contextually + locally relevant + authoritative (venn sweet spot); partial relevance still fine:
1. Chamber of Commerce (high authority + local, weak context, easy win).
2. Niche business associations, search `[city] [niche] association`, check member directories give followed links.
3. Local resource pages, `inurl:resources.html site:co.uk plumbing` or `inurl:resources [city] [niche]`; tangential businesses (wedding photographer → florists/venues/cake shops' resource pages). Verify they actually link out.
4. Guest posts on local sites, Content Explorer `title:photography site:.au` + one-page-per-domain + wordcount≥500 + DR 30-60 filter.
5. Local "best of" listicles, pitch inclusion with a non-pushy email (invite them to try the restaurant; "no obligations").
6. **Link Intersect**, domains linking to 2+ local competitors but not you = warmest prospects; more competitors linked = more relevant.
7. Also: local news/sponsorships/community events (from beginner course: sponsor local events, local PR).

## YOUTUBE SEO
**Why search views:** consistent (vs suggested/browse which need an existing audience), high AVD, compounds subs. Ahrefs: 1.3M views/yr from YT search; 172k views/yr from Google search to videos.
- YouTube algo per YouTube: "title, description, and video content match the query" + engagement signals (CTR, watch time, retention, likes/comments/shares). **Engagement is the prerequisite**, no watch time, no rankings.
- KW research: YouTube autosuggest (+ `*` wildcard), Google Trends set to "YouTube search" (relative only), TubeBuddy/vidIQ topic tabs (their volume estimates are wildly inflated, verified vs real data), Ahrefs Keywords Explorer YouTube engine (only tool w/ country-level YT data).
- **Google video rankings**: only target keywords where videos ALREADY rank in Google SERPs (video carousel present = video intent). Find topics: Content Explorer `site:youtube.com inurl:watch title:topic` sorted by organic traffic (check 3-yr consistency); Site Explorer youtube.com → organic keywords + niche terms; scrape `site:youtube.com` Google results w/ 100/page + Scraper.
- On-video checklist: exact keyword in title (thumbnail sells, title ranks); real summary description w/ keyword in first lines; timestamps→chapters (rank independently in Google "key moments"); SAY the keyword aloud (Google parses audio/transcripts); match format of ranking videos; filename/can-use-video-object-schema if embedding.
- Playlist strategy: series playlists for the "Up next" suggested slot; tight topic focus early on to build a relevant audience.
- Retention editing (10-step Ahrefs process): clean audio (Audition/Audacity noise removal), cut ruthlessly, jump cuts, b-roll to reset attention, pattern interrupts. AVD is the ranking fuel.
- Discovery ads can seed engaged views (~$0.05-0.07/engaged view, "earned actions" = real subs/views within 7d), optional accelerator, in-stream is for awareness only.


---


# Ahrefs, Content Marketing, Blogging for Business, Affiliate SEO & Misc Tutorials

> Source: Ahrefs YouTube playlists (Blogging for Business course, Content Marketing, Affiliate Marketing for Beginners, Link Building Tutorials, SEO Tips/rank1, Beginner Tutorials, WordPress). Transcripts: `D:\seo-courses\transcripts\ahrefs-*`.

---

## 1. Why blogs fail, the "spike of hope" model (Blogging for Business, Tim Soulo)

- After publishing, traffic comes from: **existing audience** (email/social, zero new people), **communities** (new people first time, decays with reuse), **word of mouth** (small unless an influencer amplifies). This produces the "spike of hope" → "flatline of nope" (Rand Fishkin’s terms).
- Viral traffic doesn’t snowball, Pulsar research shows it spreads in **decay**: needs a huge hub (influencer) to start, then dies fast. Both requirements (sticky content + influencer kickoff) are extremely hard to repeat.
- **SEO traffic is the only compounding channel**: it builds slowly but doesn’t fade. Ahrefs blog publishes ~1 post/week and grows because each post targets specific keywords and ranks.
- **Data point:** only **5.7% of newly published pages** reach Google’s top 10 for ≥1 keyword within a year (Ahrefs 2M-page study). "Publish & pray" fails; ranking requires deliberate intent-matching + backlinks.
- **"Publish more often" is a myth.** One keystone article can carry a business (Healthline’s "how to lose weight fast" ranked for 9,000+ keywords, ~300k visits/mo). Note: Ahrefs traffic estimates undercount ~2–3x.
- **Traffic potential beats search volume**: estimate a topic’s true traffic by checking how much *the current top-ranking page* gets across all its keywords, a page ranks for thousands of terms, not one.

## 2. Promotion strategy

- **Content promotion’s real job is acquiring backlinks**, not same-day traffic. Small sites can’t rely on natural links: ~1 natural link per ~1,000 visitors, people link to known/trusted sites, most niches' audiences don’t own websites.
- **The 110/110 rule:** go all-in on BOTH exceptional content AND relentless promotion, don’t stop at the checklist; promote until the page ranks.
- Promotion channels: communities (join 5–10, be genuinely valuable, mods tolerate occasional plugs), repurposing/syndication (YouTube versions can 10x reach, an Ahrefs audit video out-viewed the article 600k vs 60k), guest blogging, Quora/comments, HARO/journalists, podcasts, outreach, paid (justify ads: "free" promotion costs your hourly rate).
- **Outreach rules:** never reach out without an excuse; make the email about *them*; three working excuses, (1) fresh angle on what they care about, (2) new proof for their existing stance, (3) ego-bait (you featured them). Don’t ask for the link directly, "if you find it useful, perhaps consider linking", the content persuades, not the request. **Max one follow-up**; nag sequences get publicly ridiculed.

## 3. Product-led content (Ahrefs' own strategy)

- Integrate the product as the hero solving the searcher’s problem ("winterize sprinkler system" → needs an air compressor → recommend yours). Serves awareness, acquisition, and retention at once.
- **Business Potential score 0–3** for topic prioritization: 3 = product irreplaceable for the problem (e.g., Ahrefs for "rank higher on Google"); 2 = helps a lot, not essential; 1 = fleeting mention only; 0 = can’t mention → drop. Prioritize traffic potential × business potential.
- Works when the product solves common problems people search about; local pubs/chiropractors should use local SEO instead.

## 4. Content hubs (pillar-cluster)

- Structure: **hub/pillar page** (in-depth guide on broad topic) ↔ **cluster/subpages** (each a specific subtopic) ↔ **bidirectional internal links**.
- SEO mechanics: builds topical authority; backlinks to any hub member lift all members via internal PageRank flow. (Drift’s chatbot hub: 500+ linking domains, ~6k monthly visits in 7 months.)
- Hub criteria: ~5–20 viable subtopics; targets a *broad informational* query with real volume; top-10 SERP must show informational intent (don’t hub a navigational/transactional SERP).
- Finding subtopics: listicles on the topic ("list of yoga poses"), Wikipedia’s internal structure, keyword tools' phrase-match + questions reports.

## 5. Affiliate SEO course (unique mechanics)

- **Product cluster** = general comparison post ("best car seats") + individual product reviews + branded "X vs Y" comparisons, all interlinked; reviews link back to the comparison hub.
- **Content ratio:** roughly equal commercial vs informational (~50/50, 40/60, 60/40 all fine). All-commercial can’t earn links (Wirecutter was the exception, hands-on testing); all-informational caps revenue.
- **Templates:**
  - *Best-of listicle:* short credibility intro ("we tested 10, rated by comfort/price/safety") → "Best X at a glance" skimmer summary → consistent-format list points (H2 = product name) → brief conclusion.
  - *Product review:* short intro + internal link to the comparison hub → body = the niche’s buying categories (car seats: safety/size/comfort/install/portability/specs, same categories every review = template) → conclusion: who it’s best for.
  - *Branded comparison (X vs Y):* reader already narrowed to 2, compare category-by-category (tables), link to both reviews + the hub.
  - *Informational:* listicles (numbered benefit-driven H2s, TOC jump links, mini-template per item: why/how/action), step-by-step guides (**PSP intro** = Problem→Solution→Proof; H2 = Step N), expanded-definition posts ("come for the definition, stay for the details"; subtopics from PAA + related-terms reports).
- **Intent check is mandatory before writing**, "how to clean car seats" means car seats in a car, not child seats; the SERP’s top pages reveal the format to copy (e.g., tips-listicle beating step-by-step).

## 6. Internal linking (standalone tutorial)

- Three free discovery methods: (1) `site:domain "topic"` + Ctrl-F for unlinked mentions; (2) Site Audit link-opportunities reports (semantic, excludes already-linked); (3) **Best-by-links power pages**, route authority from your strongest pages to targets that need it.
- Rules: logical hierarchy (homepage → topics → subtopics), content hubs for depth, descriptive anchors, don’t force links between irrelevant pages, don’t overdo it.

## 7. Link-building tutorial extras

- **First 100 backlinks / new sites:** start with tactics needing no existing authority, reclaim unlinked mentions, cite in relevant existing articles, directories/communities, guest posts, competitor backlink replication.
- **Reciprocal links study:** 73.6% of domains have ≥1 naturally reciprocal link; ~43.7% of top-10 pages have some reciprocity, it’s a natural web byproduct. Verdict: don’t run link exchanges (against guidelines), but don’t avoid linking to sites that link to you if they genuinely fit. Build *relationships* with people you already cite repeatedly (Linked-domains report → intro email → give value first).
- **Resource-page link building:** find `[topic] + "resources"/"useful links"` pages, pitch only genuinely link-worthy assets.
- **Guest posting at scale:** prospect via content-explorer + Sheets; personalize the opening only; one follow-up.
- **Broken link building:** still works, find dead pages with backlinks (esp. competitors' 404s), recreate equivalent content, pitch replacements.
- **Negotiation:** links are relationships, lead with value, make the ask easy and specific (exact placement suggestion).

## 8. Quick wins from SEO-tips & beginner tutorials

- **Title tags for CTR:** include the query’s core promise + specificity (numbers, year, benefit); front-load keywords; match what top results promise but differentiate; ~50–60 chars.
- **Featured snippets:** find keywords where you rank top-10 and a snippet exists (Ahrefs SERP-features filter); match the winning format (paragraph ~40–60 words, list, or table); place the answer directly under an H2 phrased as the question.
- **5 things that don’t matter much:** exact-match domains, keyword density targets, meta keywords, excessive URL parameters panic, obsessing over a single ranking factor.
- **What matters (consensus):** intent-matched content quality, backlinks, technical crawlability, internal linking, CTR/title appeal.
- **Ranking ≠ traffic:** #1 on a low-volume multi-intent keyword can underperform #3 on a keyword whose SERP features leave clicks; judge keywords by *traffic potential + business value*, not position.
- **Content audit playbook:** pull GSC + crawler data → for each page decide keep/improve/consolidate/delete; refresh declining pages; merge thin overlapping pages into a stronger canonical target.
- **How long to rank:** data shows top-10 pages are usually 2+ years old; new pages that break in tend to be low-competition, front-load those, build authority for the rest.
- **WordPress speed:** caching plugin, image compression/WebP, lightweight theme, limit plugins, CDN; test with PageSpeed Insights field data not just Lighthouse.
- **ChatGPT for SEO, good uses:** ideation, outlines, content refresh drafts, schema generation, title/meta variants, entity suggestions. **Bad uses:** unedited publish-at-scale, factual claims without verification, expecting it to know search volumes.


---


# COURSE: Ahrefs AI SEO / AEO Course (Sam Oh), 12 lessons, Dec 2025
Source: https://www.youtube.com/playlist?list=PLvJ_dXFSpd2tU_vPcmzqP8L3qVPDoM9Rf | Transcripts: D:\seo-courses\transcripts\ahrefs-ai-seo-course\
(Related older written course on Desktop: Ahrefs-AEO-Course.md, same concepts, this is the newer video version.)

## The case for AEO (with data)
- AI Overview on a SERP cuts CTR of the #1 page by **58%** (Dec 2025).
- ChatGPT: 900M weekly users, ~12% of Google search volume. AI referral traffic grew **9.7x YoY**; ChatGPT alone +85% since Jan, now sends more traffic than Reddit or LinkedIn.
- AI traffic ≈ only **0.25% of total site traffic** on average; Google still sends ~210x more than all AI platforms combined. BUT: Ahrefs AI visitors convert **23x** organic; Vercel sees 10% conversion; Tally says AI is their largest acquisition channel (+$1M ARR). AI-referred visitors are pre-qualified.
- AEO = Answer Engine Optimization (aka GEO/LLMO): getting mentioned/cited inside AI-generated answers (AI Overviews, AI Mode, ChatGPT, Perplexity, Copilot, Gemini). No list, no positions, you compete for a *mention*. SEO is the foundation; AEO is a layer on top, not a replacement.

## M1, How AI search works
**Two information sources:** (1) **Training data**, static snapshot of the web, refreshed ~6-monthly; being mentioned widely enough bakes you in. (2) **Real-time retrieval (RAG)**, AI searches the web live for fresh/specific questions → classic SEO skills directly control this.

**Query fanout**, THE key mechanic: one prompt expands into many sub-queries (avg **9–11**, up to 28; Deep Research ran 420 for one prompt). Content must cover the *whole topic*, not one keyword, if your "start a podcast" page skips equipment/hosting/promotion, AI cites someone else’s. Fanout queries are visible in Brand Radar but are synthetic, inconsistent, 95%+ zero-volume → treat as a *topic map*, not a keyword list.

**Citations are probabilistic**, same question 5x → cited maybe 3/5. What raises the odds: **consensus** (multiple sources agreeing about you), **freshness** (AI-cited content is ~25% fresher than SERP content), **authority** (76%→now ~38% of AIO citations from top-10 pages; 14% of AIO-cited pages don’t rank top-100 at all).

**Platform differences (cite different sources, only 14% overlap of top-50 cited domains across AIO/ChatGPT/Perplexity):**
- **AI Overviews**: authoritative established sites, health/finance/encyclopedic, YouTube (~5.6% of citations), Reddit.
- **AI Mode**: YouTube #1 by far, Google, Wikipedia; cites Quora 3.5x more than AIO; heavy on social (FB/IG). Only **13.7% citation overlap with AIO despite 86% semantic answer similarity**.
- **ChatGPT**: publishers/media (Reddit, Wikipedia, Amazon, Forbes, BI, Wired; median DR of cited pages = 90, partly licensing deals). YouTube mentions have **0.737 correlation** with ChatGPT visibility, strongest factor studied.
- **Perplexity**: most aligned with Google (28.6% of citations from Google top-10 vs ChatGPT’s ~8-10%).
- Prioritize: Google AI + ChatGPT = vast majority of volume. If already ranking → AIO/Perplexity head start; ChatGPT → get on high-DR editorial sites.

**3 visibility types:** cited+linked (best, rarest, only **~28% of mentions include a link**: Perplexity 51.6%, AI Mode 36.8%, ChatGPT 26.9%, AIO 10.7%); mentioned-not-linked (word-of-mouth at scale → drives branded searches; feeds training data, branded web mentions have **0.664 correlation** with AIO visibility, stronger than backlinks/DR/RDs); invisible (find and fix). Weighted by impressions, links appear on the highest-volume queries (Perplexity links in 78% of impressions; Gemini 71% of impressions vs 16.8% of mentions).

## M2, Strategy
**Brand gap analysis:** (1) Map branded entities, main brand, sub-brands, product names, proprietary metrics (e.g. "Domain Rating"), personal brands; attach each to the topics/attributes it *should* be known for (use keyword research for recurring modifiers: affordable, AI-powered, enterprise…). (2) Audit mentions/citations/impressions/AI-share-of-voice per platform (Brand Radar); filter for "responses mention brand but don’t cite site" (= missed citation opportunities) and "competitor mentioned, I’m not". (3) Classify gaps: **visibility** (appearing less than competitors), **narrative** (AI describes you wrong, e.g. "budget" when you’re premium), **topic** (missing associations), **format** (no listicles/videos/comparisons), **web mentions** (not on third-party listicles/reviews/forums), **demand** (branded queries you don’t capture).

**Keyword research for AEO:** generate from broad seeds (Matching Terms); vet with **BID** = Business potential (would #1 actually help?), Intent (SERP match), Difficulty (RDs/DR of ranking pages; low-DR sites in top 10 = good sign). Then the **AI filter**: can AI fully satisfy the query? AIOs appear on ~21% of keywords but ~58% of question queries, 46% of 7+-word queries; 99.9% of AIO keywords are informational. If AIO nails it → target for *mention* not click. **AI-proof categories**: free tools/calculators/generators/templates/checkers (people must DO something, no AIO); transactional intent queries.

**Find mention targets:** Brand Radar → "others only" (competitors mentioned, you’re not) → filter queries containing best/top/versus/review/alternative. 43.8% of all AI-cited pages are **listicles** (they build consensus). >45% of AIO citations change per refresh (~every 2 days) → re-check regularly.

**Prompt research:** users write full-context natural language; same question 10 ways → 10 answers. Don’t optimize for a prompt, build visibility across the whole topic.

## M3, Execution
**Content that gets cited (data):** length irrelevant (correlation 0.04; 53.4% of cited pages <1,000 words). Freshness critical, cited content 25.7% fresher; 89.7% of ChatGPT’s top-cited pages updated in 2025, 76% within last 30 days, make *meaningful* updates, not date changes. Formats: listicles/comparisons/reviews (43.8%), original data/stats, X-vs-Y pages.

**Writing principles:** **BLUF**, answer first (LLMs weight beginning/end of passages like human F-pattern). **Atomic content**, every H2 section must make sense standalone (AI chunks at arbitrary boundaries). **Entity-rich**, name brands/products/people/numbers, not "this tool". **Simple declarative**, short SVO sentences, one idea each. **Brand-label your frameworks**, "the Ahrefs content scoring matrix" not generic, then distribute widely, or AI absorbs the idea without credit. **Refresh sleeper pages**, Top Pages → sort by traffic decline → pages with links + declining traffic = fastest AEO wins (verify decline is content-staleness, not lost links).

**Earning mentions (3 tiers):** T1 = third-party editorial (publications, Wirecutter-style reviews, listicles, YouTube reviews), hardest, most valuable (mentions on highly-linked pages: **0.7 correlation** with AIO). Find via Cited Domains report + Content Explorer `title:best|versus -yourbrand` filtered by RDs. T2 = UGC/community (Reddit is a top ChatGPT source + LLM training source), answer real threads, never spam; find ranking Reddit threads via Site Explorer reddit.com → organic keywords top-5 + niche terms. T3 = own properties (YouTube, podcast, LinkedIn, all indexable citation sources). Run periodic **mention audits**, track drops, fix misinformation fast (update your content, then request publisher correction).

**YouTube for AI:** most-cited domain in AIO; GPT-4 trained on 1M+ hours of transcripts. Target *search hits* not viral hits, find topics via Site Explorer → youtube.com/watch → organic keywords top-3 + niche terms. Checklist: keyword literally in title; real description summary w/ keyword early; timestamps→chapters (rank as their own snippets); **say the keyword aloud** (Google understands audio); match format of what’s ranking. Then check Brand Radar topics report filtered to youtube.com mentions for AI-specific angles.

**Technical AEO (6 checks):** (1) **robots.txt**, don’t block GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, Google-Extended; ~5.9% of 140M sites block GPTBot, often accidentally (Cloudflare’s "instruct AI bot traffic" is ON by default). (2) **JavaScript rendering**, ChatGPT’s crawler does NOT render JS (Gemini/Copilot do); need SSR/pre-render; test by disabling JS. (3) **Page speed**, real-time retrieval drops slow pages before scoring. (4) Clean HTML hierarchy (H1→H2→H3, self-contained sections). (5) Schema, no proven AEO boost, doesn’t hurt, keep for SEO. (6) **Hallucinated URLs**, AI 404s are 2.87x Google’s rate (ChatGPT ~1% of clicked URLs); watch AI-referrer traffic hitting 404s → 301 to nearest real page. `llms.txt`: no major provider officially supports it, optional, low priority.

## M4, Measurement & action plan
**3 measurement pillars:**
1. **AI referral traffic**, GA4: Admin→Data display→Channel groups→copy default→add "AI traffic" channel, source regex `chat.openai.com|perplexity|gemini.google.com|copilot.microsoft.com|claude.ai|deepseek.com`. CAVEAT, undercounts: ChatGPT in-content paid links send no-referrer; Perplexity desktop app, Copilot on Windows, Grok = invisible; shows as direct. Ahrefs Web Analytics has a built-in AI channel + separates unknown from direct. Watch: pages getting AI traffic (keep fresh + CTAs) and important pages NOT getting it.
2. **AI bot activity**, training bots (GPTBot, Google-Extended) vs search/citation bots (ChatGPT-User, OAI-SearchBot, these drive referral traffic). Server logs or Ahrefs Bot Analytics (free Cloudflare plan works). Heavy bot hits on a page = likely citation source; important pages with zero bot visits = discoverability/internal-linking problem.
3. **Self-reported attribution**, add "How did you hear about us?" (options: AI assistant/ChatGPT/Perplexity/AI search) to signup/checkout/surveys. Ahrefs: ~3% of conversions self-report as AI. Most AI influence appears as "direct" or branded organic, asking is the only way to see it.

**Misinformation risk (tested):** fake brand + 3 contradicting planted sources → Gemini & Perplexity repeated falsehoods in **37%** of answers; ChatGPT stayed <7% and cited the official FAQ 84%. Defense: publish specific official FAQ (numbers/dates/facts, AI picks "specific fiction over vague truth"), monitor sentiment monthly, correct third-party sources fast.

**First-week action plan:** check robots.txt for AI blocks (5 min) → set up AI analytics + self-attribution question → meaningfully refresh top 5–10 pages → run brand gap analysis → list top-10 mention targets (cited domains/pages reports). Ongoing: monthly Brand Radar check, quarterly competitive audit.


---


# COURSES: HubSpot AEO (6 vids) + Semrush LLM Visibility (9 vids)
Sources: https://www.youtube.com/playlist?list=PLU8vC1IJrS7o3eqYpMPpKvVKdC4bD9jRp (HubSpot AEO), Semrush LLM Visibility playlist | Transcripts: D:\seo-courses\transcripts\hubspot-aeo\, semrush-llm-visibility\
Builds on ahrefs-ai-seo-course.md, this file captures the DELTA (unique data & tactics not duplicated there).

## Unique data points (Semrush studies)
- **88% of cited pages are unique to ONE platform**, 4 platforms, same query, nearly disjoint source pools. Citation overlap ChatGPT-vs-Perplexity test: 1 URL shared out of ~36.
- **Only 44% of Google’s top-10 pages appear in ANY AI answer.** Per-platform overlap with Google top-10: Perplexity 32%, AI Mode ~15%, AI Overviews ~8%, **ChatGPT just 2%**. → Google ranking ≠ AI visibility; ChatGPT is its own game.
- **Platform index dependencies:** ChatGPT leans on Bing’s index (+ some Google/own crawler evidence), Google AI products use Google’s index, Claude uses Brave Search, Perplexity runs its own crawler+index. → **Bing Webmaster Tools indexing matters for ChatGPT visibility**, most people never check it.
- BuzzSumo: original research earns **3x more citations** than other content types.
- Gartner prediction cited by HubSpot: ~25% of traditional search traffic displaced by AI answers by 2026.
- 56% of consumers use AI tools to compare products/decide purchases (industry study cited by Brian Dean).

## HubSpot GEO 7-step system (video 006) + 5-free-tools stack (001)
1. **Audit**, take top-10 money keywords → ask AI as customers would (private/incognito chat so results aren’t biased): "What’s the best X?", "How do I choose X?" across ChatGPT/Perplexity/AIO. Log which competitors appear + which sources get cited (G2, Reddit, comparison pages recur). Free graders: HubSpot AEO Grader scores Brand recognition / Market position / Presence quality / Sentiment / Share-of-voice across ChatGPT+Perplexity+Gemini in 30s. Screenshot = baseline; re-run monthly. Tools mentioned: Profound, Aerops, Semrush AI Visibility Toolkit.
2. **Restructure**, answer-first ordering (flip story→answer to answer→story); use the exact modifiers buyers type into AI ("best", "top", "cheapest", "easy to use"); ship listicles + comparison pages even if self-ranking feels awkward, AI reads it as structured trustworthy data.
3. **Authority**, schema via Google’s Structured Data Markup Helper → JSON-LD → validate in Rich Results Test; external validation on G2/Capterra/Trustpilot/Yelp; publish original research; journalist outreach (HARO-style).
4. **Technical**, <2s load, mobile-friendly, ordered headers, clean URLs, unblocked AI crawlers, indexable on Google AND Bing.
5-7. (Ongoing) monitor citations monthly, "cited once isn’t forever, AI re-checks sources"; track category dips vs. your dips (AI search sensor / Semrush Sensor); build a persistent "AEO assistant" chat = feed your grader scores + AEO report into one AI chat, rename it, run every new page through it before publishing.
- **Presence vs reputation:** decent sentiment + low visibility = presence problem (fixable); high mentions + wrong facts = reputation problem (fix sources first).
- Mike King (iPullRank) on HubSpot pod: **meta descriptions matter MORE for ChatGPT**, unlike Google SEO where omitting them boosts CTR, LLMs want the answer spoiled in the meta. Reddit is "working like gangbusters"; own-subreddit engagement correlates with positive brand mentions across ALL of Reddit. For links: **5-10 highly relevant links > 1,000 mediocre**, relevance over volume now.

## Semrush, the "citation core" concept (video 001, Chirp case)
- Winning = being mentioned favorably AND cited. A product with CNN/Elle/Lifehacker coverage still lost because ChatGPT’s answer drew 100% from **ranked listicles**, the coverage was in the wrong SOURCE TYPES. Lesson: map your **citation core** = list the prompts you want to win → run them → note the exact source types/pages each LLM pulls from → get into THOSE pages (not just "big publications").
- Product pages need explicit use-case + differentiation language, a golf magazine out-cited the brand for its own product because Chirp never stated its positioning; "spinal decompression" was hidden behind a click.
- B2B site checklist: **public pricing** (not "contact sales", vague pricing makes AI recommend competitors), clear ICP use-cases on product pages.

## Semrush, AI Share of Voice + attribution gap (002, 003)
- AI SoV = your mentions ÷ total category mentions ×100. Track per platform, even SoV on 3 platforms + near-zero on 1 = platform-specific issue worth investigating.
- 3-tier attribution framework: T1 **eligibility** (robots.txt allows GPTBot/ClaudeBot/PerplexityBot; extractable structure; indexed on Google+Bing, check GSC Pages report AND Bing Webmaster). T2 **directional metrics** (AI SoV, referral traffic, branded-search lift, cross-reference movements, none is definitive alone). T3 **business outcomes** (conversions/self-reported attribution).
- The invisible path: user asks ChatGPT → Googles your brand → converts. Analytics credits organic; AI did the work. Self-attribution question is the fix.

## Semrush, brand playbook case studies (004)
- **Tally**, founder answered Reddit questions for YEARS (help-first, rare product mentions) → ChatGPT = #1 signup source, 2,000+ users/week, AI = largest acquisition channel. Zoho does it at enterprise scale.
- **Patagonia**, policies that generate UGC (Ironclad guarantee → customers tell return stories unprompted; Worn Wear; public stances) → highest AI SoV in fashion vertical. Design talking points INTO the product/service.
- **Garmin**, press kit page (specs/pricing/hi-res images ready for journalists), expands into adjacent listicle categories (running/cycling/multisport), times pitches before annual "best-of" refresh cycles → consensus of identical specs across dozens of publications = AI treats it as verified.
- **SentinelOne**, proprietary threat research nobody else has = exclusive citation; research → Forbes/TechCrunch coverage → compounding.

## Semrush, Rita Cidre: 4 reasons brands are invisible (006)
1. **Reviews are wrong-type**: AI reads review TEXT not star ratings, "great product" gives nothing to cite. Fix: prompt customers with specific questions ("How did our dashboard compare to your old tool?"), diversify platforms (G2/Capterra/Trustpilot/Amazon).
2. **Absent from conversations**: monitor brand+competitors+category on Reddit/Quora/forums (brand monitoring alerts); answer helpfully, respond constructively to complaints, a good public complaint-response is a stronger trust signal than 5-star walls.
3. **Only you talk about you**: micro-influencer detailed content ("exactly how I set it up") > sponsored reach; give customers a shareable results framework ("before X, after Y in Z timeframe").
4. **Site doesn’t feed AI**: B2B → public pricing, comparisons; structure for extraction.

## Brian Dean × Semrush (005, 008)
- Strategy: prioritize **buying keywords** (alternatives/comparisons/pricing/reviews), informational queries are AI-answered and low-click; buyer-intent queries still produce clicks+leads in both channels.
- **Chunking pattern**: descriptive question-like subheading (could be an article title itself) → 1-2 sentence direct answer immediately → proof (data/screenshots/pricing/pros-cons/expert input). Treat each section as a mini-article.
- **Entity authority**: state who you are/what you do/who you serve identically on About + product pages + homepage, repetition across your own site builds AI confidence.
- Consensus is the algorithm: Keychron/Caraway win because review sites + Amazon reviews + Reddit threads + YouTube roundups all describe them the same way. Steps: join product discussions honestly (even say when you’re not the best fit) → get into *specific-problem* best-of lists ("best X for sensitive skin" not "best skincare") via short outreach + free product → build your own comparison/alternatives pages (competitor "alternatives" pages get cited AND linked) → maximize the sentence around every mention (specific audience+benefit, not "high quality").
- **12 brand authority signals** (009): T1 weekly, branded search volume (cleanest demand signal; unbranded growth without branded growth = forgettable), co-occurrence queries ("Zendesk for small business", GSC regex filter excluding pure-brand queries), navigational intent ("brand login/pricing/contact"). T2 monthly, backlink QUALITY (1 editorial link > 1000 directories; over-SEO detector), review velocity+sentiment, media mentions, NAP/brand-identity consistency everywhere ("AI amplifies what’s there, without clear brand it scales mediocrity"), E-E-A-T signals (author bios, Person schema w/ knowsAbout, case studies, methodology, Trust = foundation per QRG), topical authority = DEPTH not breadth (Kevin Indig "topic share"; HubSpot lost traffic straying from core topics).


---


# Semrush, SEO Courses, New-Era SEO & Audits

> Sources: `semrush-new-era-seo`, `semrush-2025-strategies`, `semrush-seo-essentials`, `semrush-seo-quickstart`, `semrush-free-seo-training`, `semrush-serp-mastery`, `semrush-backlinks`, `semrush-local-seo`, `semrush-content-marketing`, `semrush-competitor-research`, `semrush-all-things-google`, `semrush-how-content-works`, `semrush-practical-seo-advanced`. (LLM-visibility playlist is covered in `aeo-hubspot-semrush.md`.) Transcripts: `D:\seo-courses\transcripts\semrush-*`.

---

## 1. Search Everywhere Optimization (Brian Dean / Exploding Topics, the flagship 2025+ framework)

"Search is a behavior, not a channel" (Rand Fishkin). Buyers now search across Google + Amazon + YouTube + Reddit + AI tools in one journey. Strategy:

1. **Learn who searches & how they found you**, ask existing customers directly; expect surprises (Exploding Topics discovered investors searching "alternative data providers").
2. **Map the journey in 3 stages:** Discovery (problem-aware, educational content) → Comparison (evaluating options, "X vs Y", "best tools for Z", reviews, Reddit threads) → Action (pricing, demos, case studies). AI tools are used at *all* stages.
3. **Pick the few platforms that matter**, you’re not trying to be on 50 platforms; find the 3–5 where your audience actually searches (for them it was YouTube + Substack newsletters, the newsletter grew to 155k subs ≈ 25% of ARR).
4. **Create stage-matched content per platform:** educational for discovery, comparison/category pages for evaluation, proof/case studies for action.

## 2. AI brand-mention playbook (extends the AEO section)

- **Consistent wording is the mechanism:** LLMs connect your brand to a problem when the *same description* appears across many sources. Reuse one clean positioning phrase everywhere, your site, outreach, Reddit answers, press releases.
- **Get into existing "best of / comparison / vs" pages**, they rank, get traffic, and are the pages AI systems cite most. Find them by prompting LLMs with best-of questions and noting cited sources. Pitch short + factual + hype-free; goal is the *description*, not even the link (nofollow/unlinked still counts).
- **Replacement wedge:** flag dead/deprecated tools on listicles (e.g., Google Correlate shutdown) and suggest yours as the updated entry.
- **Communities:** answer only high-intent threads ("what’s the best tool for X", "alternatives to Y"), grounded and non-salesy, same phrasing each time.
- **Press releases:** weak as PR, valuable as a canonical neutral description that then propagates across channels.
- **Build a content machine (the #1 factor, echoes Mike King: "AI search is a content strategy problem, not a technical one"):** free no-login tool tier, weekly newsletter, YouTube channel; create once → redistribute to 5–50 pieces; brand mentions then accrue automatically 24/7.
- `llms.txt` and content chunking are marginal next to genuine brand chatter.

## 3. Modern SERP/algorithm understanding (SERP Mastery + 2025 strategies)

- RankBrain-era reality: engines match **intent and context**, not keyword strings, write for the *job to be done*, not the term.
- **Zero-click is an opportunity**: own snippets/SERP features even when users don’t click, brand visibility compounds.
- Engagement signals aren’t "official" ranking factors but are quality proxies: dwell time, CTR, pages/session, optimize for humans and the proxies follow.
- E-E-A-T = Experience, Expertise, Authoritativeness, Trust: demonstrate via author bios, first-hand evidence (photos/data), citations, transparent about/contact pages, and third-party corroboration.

## 4. Site-cleanup list ("9 things to remove", 2025)

- **Blog dates:** keep them, but actually update the content, undated OR outdated both erode trust.
- **PDFs:** can’t optimize meta/nav, static across devices, convert to HTML pages or gate behind email signup.
- **Dead-end thank-you pages:** add related products/content/email CTA, keep momentum.
- **Popups:** only for high-value actions, delayed or scroll-triggered, trivially closable, anything interrupting the experience hurts SEO.
- **Redirect chains/loops:** Google tolerates ~10 hops but they slow pages and dilute equity; redirect to the *closest intent match*, audit with Site Audit.
- **Clashing colors / vague headlines / bloated nav / long forms:** ≤2 accent colors; headlines that state the offer + keywords; nav ≤5 top items (rest to footer); forms short early in the journey, attachments for complex inquiries.

## 5. The Semrush-style audit workflow (from free-training + quickstart)

Order of operations: **keyword/intent audit → technical → content → links**.

- **Intent mismatch hunt:** check which pages rank for which intent. Classic find: a transactional page trying to rank for informational queries (e.g., lawyer’s service page ranking for "statute of limitations" terms) → build *separate* informational assets for those terms, internally link them to feed the transactional page. Geo-targeted informational queries are "golden opportunities" for local topical authority.
- **Technical priorities:** crawlability > indexation > everything. Crawl depth: **no page >3 clicks from homepage** on small sites. Deep pages signal low importance to Google AND often correlate with thin content, fix by restructuring OR deleting.
- **Keyword cannibalization via URL structure:** don’t nest children under a folder that repeats the parent’s keyword (`/st-louis-personal-injury/wrongful-death` competes with `/st-louis-personal-injury`). One page per primary keyword; consolidate duplicates. Delete tag archives on small sites, they only create duplicate pages.
- **Perspective:** micro technical fixes rarely move the needle, the levers are right keywords → best page for intent → topical relevance → links. Don’t burn sprint time on warning-level trivia while those four are undone.
- **Content audit:** connect GA + GSC → bucket pages rewrite/remove/update/keep; low avg-view-duration + declining engagement = refresh candidates; prioritize pages with existing traffic/backlinks.

## 6. Link-prospect vetting, the PBN detection checklist (Semrush Weekly Wisdom)

Before accepting/buying any link placement:

1. `site:domain`, is it even indexed, and at plausible volume?
2. Nav full of "Submit guest post / Write for us" → content farm built to sell links.
3. **Wayback Machine**, recent ownership change or expired-domain flip (was a personal football blog, now a "home improvement" site) → PBN.
4. **Reverse-IP lookup**, shared with pills/porn/casino domains or 1,000+ junk sites → bad neighborhood.
5. **Stock-photo author bylines** reused across many sites → lazy PBN footprint.
6. Traffic check, near-zero organic clicks or a cliff from a past quality update confirms it.

Legit markers: real-person author with a real footprint elsewhere, topical posts not built around guest-post pitches, reasonable shared-hosting neighborhood, actual organic traffic.

## 7. Misc operational guidance

- **Local SEO (80-min course + GBP videos):** claim/verify GBP; primary + secondary categories; complete every field (services, hours, attributes); photos with real imagery; weekly posts; Q&A seeding; review velocity + responses; NAP consistency; embed map + local schema on the site.
- **New website first-30-days:** GSC + GA4 + sitemap on day 1; keyword map before writing; 5–10 intent-matched core pages; internal links; 2–3 early authority links (profiles, local citations, niche directories).
- **Competitor research:** traffic analytics for channel mix; keyword gap for terms they rank and you don’t; backlink gap for domains linking to 2+ competitors but not you (warmest prospects); top pages by links/traffic as content targets.
- **GA4 essentials:** events > pageviews, mark conversions (form submit, signup, purchase); Exploration reports for funnels; compare segments (organic vs others); connect to GSC for query data.
- **Conversion basics:** single CTA per page; above-fold value prop; social proof near CTAs; reduce form fields; speed = conversion.
- **YouTube SEO (Semrush):** keyword in title/description/chapters; retention-first editing; consistent series; CTR via thumbnail+title pairing.
- **Hiring/agency lessons:** SEO sits at the table when reports speak revenue (pipeline, CAC, LTV) not positions; scope retainers on outcomes not deliverable counts.


---


# Moz, Whiteboard Friday, One-Hour Guide, Technical Tips

> Sources: `moz-whiteboard-friday` (AI search episodes, index bloat, clusters, authority), `moz-one-hour-guide` (canonical 6-part SEO framework), `moz-technical-tips`, `moz-daily-seo-fix` (Moz Pro tool usage). Transcripts: `D:\seo-courses\transcripts\moz-*`.

---

## 1. The One-Hour Guide framework (Rand Fishkin’s canonical model)

SEO strategy in six parts, strategy → keyword research → searcher satisfaction → on-page → technical → links:

- **Strategy first:** define what ranking accomplishes for the business *before* keyword work, rankings that can’t produce customers are vanity.
- **Keyword research:** map searcher language, not internal jargon; group by intent; prioritize by value × difficulty.
- **Searcher satisfaction is the deeper goal:** Google’s real job is answering the query well, satisfy the *reason* behind the search, not just the literal keywords.
- **On-page:** smart keyword targeting (title, headline, first paragraph, relevant entities), unique value over comprehensiveness theater.
- **Technical:** crawlable, fast, mobile-friendly, secure, the table stakes layer.
- **Link building:** earn through useful things people want to cite; links remain the clearest authority signal.

## 2. AI search episodes (2024–2025 Whiteboard Fridays, the freshest Moz thinking)

### AI Mode & query fan-out (Tom Capper)
- **AI Mode = separate Google tab** with conversational, ChatGPT-like format. Uses **query fan-out**: one prompt triggers dozens of related searches whose results are merged with citations.
- Not broken out in Search Console, AI Mode traffic is lumped into regular "web" search data.
- Citations exist but expected to drive few clicks. Accuracy is mixed, test it on topics you know deeply.
- Early data (mid-2025): AI Mode usage trailed even News/Images tabs; Sundar hinted it’s the future of search but Google itself was still testing demand. Strategic read: web search may get pushed back toward "find this specific website" while AI handles research.

### The Great Decoupling of clicks and impressions (Tom Capper)
- AI Overview prevalence grew ~50% US / doubled UK / 0→14% Ireland between Dec 2024 → May 2025, impressions rise while clicks stay flat → the clicks/impressions decoupling every site sees in GSC.
- Google’s claim AIOs would drive more clicks hasn’t matched most sites' experience.
- **Reframe: impressions are a legitimate KPI.** Other channels pay fortunes for zero-click visibility (billboards, TV). An AIO citation is brand exposure, undervalued if you only count clicks. Track impressions + brand-search volume as the visibility layer; clicks as the traffic layer.

### What changed in AI search / SEO 2021–2025 (Sussman, Shepard revisits)
- The fundamentals held: intent, quality, authority. What changed: distribution (more zero-click, more AI answers), measurement (impressions & citations alongside clicks), and the surface area (multi-platform "search everywhere").

### Britney Muller, future of AI in search (54-min interview, key takeaways)
- "AI" is not one thing, LLMs are one model class; most AI announcements are PR theater. Treat vendor claims with suspicion; evaluate *provable value per task*.
- **Human-in-the-loop is the quality differentiator**: LLM drafts + human refinement/experience >> raw output. The "growing mountain of AI slop" is a differentiation opportunity.
- Prompt quality = clarity of thought: give the model explicit goals + examples of good prior work.
- What AI can’t replicate: real quotes, first-hand stories, proprietary data, genuine opinions, invest there.

## 3. Index bloat (Tom Capper, practical fix workflow)

URL funnel: **all URLs → discovered → crawled → indexed → pages with non-trivial traffic**. Big gap between *indexed* and *traffic-bearing* = index bloat.

- Why it hurts: mass zero-traffic pages send a site-level quality signal; PageRank dilutes across dead URLs; often symptom of cannibalization.
- Common causes: unfiltered blog/UGC (staff announcements, press releases), transient listings/products/jobs, near-identical long-tail product pages.
- Fix workflow:
  1. Identify pages under ~1 click/month (check ALL channels first, don’t kill pages email/social depend on).
  2. Rescue the improvable: historically strong, backlinked, or technically blocked pages.
  3. Consolidate the rest: canonical (page must stay live) or 301 (redundant) to the *closest intent match*, consolidate best content into the target; the target can be a NEW purpose-built page.
  4. 404/noindex only for truly redundant intent-less pages, last resort: Google eventually drops noindexed URLs from crawling, and equity dies with 404s.

## 4. Striking-distance keywords (Jo Cameron)

- Keywords ranking **positions 4–20** = lowest-effort wins (already indexed + partially authoritative). Filter in rank tracker → label "striking distance" → monitor delta.
- Prioritize improvements: refresh content depth, add missing subtopics, strengthen internal links from authority pages, improve title/CTR, add supporting cluster pages.
- Layer alongside: keyword gaps (not ranking at all), protected keywords (defend top-3), competitor movements.

## 5. Topic-cluster research process (Chima Mmeje, comprehensive version)

1. **Cluster-scoped content audit** (not site-wide): keep = performing content; consolidate = cannibalizing pieces; redirect = zero-click/outdated; update = page-2/3 with potential; create = gaps.
2. **Competitive research**, export keywords competitors rank for within the cluster topic.
3. **Beyond SEO tools**, Twitter/LinkedIn/Reddit/Quora for the real questions people ask; **sales team’s call questions** for bottom-funnel topics.
4. **Modifier filtering**, "how to/what is" = top funnel, "best/vs/pricing/buy" = bottom.
5. **Cluster keywords by intent** to pre-empt cannibalization; map to journey stages.
6. **Write bottom-up**, bottom-funnel first: lower competition + higher purchase intent = fast returns that justify the cluster build.
7. **While writing:** social proof as image-CTAs, show the product in action, interlink everything ("content in isolation dies"), first-hand experience as moat.
8. **After launch:** AI outlines + human writers to scale; measure against cluster goals; **refresh top performers yearly**; social posting + influencer co-creation to reinforce topical relevance.

## 6. One domain vs many (Peter Richman)

- **SEO logic: consolidate.** One domain pools link equity → higher authority → faster ranking for competitive terms; cheaper to host/maintain.
- **Marketing logic may override:** genuinely divergent audiences/messages can justify separate sites (conversion clarity beats SEO efficiency).
- Anti-pattern: multiple domains kept because "that’s how it grew" or different CMS preferences, tail wagging the dog. Decide on audience clarity + resource pooling, not convenience.

## 7. Technical tips quickies

- **Sitemaps:** effective sitemaps = canonical URLs only, no redirects/404s, auto-generated, sectioned for index tracking.
- **Isolating 404s:** crawl + GSC + log analysis; fix internal links pointing to them first, then decide redirect vs let-die.
- **Redirect chains:** audit and collapse to single hops (A→C not A→B→C); update internal links to final destination.
- **Critical crawler issues:** prioritize by page value × error severity; monitor crawl stats after deploys.
- **Measuring technical impact:** annotate changes; compare rankings/traffic per affected template, not site-wide averages.

## 8. Other WBF one-liners worth keeping

- **Behavioral data for search:** mine on-site search queries + GA4 engagement for content gaps, users tell you what’s missing.
- **Brand authority building:** consistent positioning across PR, social, communities; authority = repeated corroboration, not self-declaration.
- **SaaS landing pages:** feature-led pages per use-case/segment outperform generic homepages; match page to intent stage.
- **Backward design for content:** start from the outcome/action you want the reader to take, design the content path backwards to the hook.
- **Community → SEO:** communities generate branded search, UGC discussions AI cites, and mention ecosystems, participate genuinely.
- **Site migrations (client comms):** set expectations, temporary ranking turbulence is normal; stage the migration; monitor redirect coverage daily.
- **When to do SEO:** earlier than feels necessary, content + authority need months to compound; bolt-on SEO after launch is a debt.


---


# Matt Diggity, Link Velocity, Anchor Text, Topical Authority, Case Studies

> Sources: `diggity-backlinks-offsite` (incl. 3 long-form webinars), `diggity-advanced-seo` (incl. Koray Tugberk topical-authority interview), `diggity-case-studies`, `diggity-seo-beginner-guides`, `diggity-website-audits`. Transcripts: `D:\seo-courses\transcripts\diggity-*`.
> Caveat: Diggity is a practitioner who openly uses grey-hat tactics (PBNs mentioned in passing). Take his *measurement frameworks* seriously; treat manipulation tactics with judgment, Google’s guidance takes precedence.

---

## 1. Link velocity, how fast to build links

- **Definition:** referring domains gained per month (not total links, one domain can link multiple times).
- **Three failure modes:**
  1. *Too fast* → sandbox that never ends, stuck rankings, manual/algorithmic penalty. His own 2017 site: ~100 links in first 2 months → quick spike to ~1k visitors/mo → steady decline → later manual penalty.
  2. *Too slow* → never close the gap. To pass a competitor you must beat their **quantity AND velocity**, they’re building too.
  3. *Inefficient* → diminishing returns. Controlled test: 1.42× velocity produced only marginally faster rankings. Find the efficient sweet spot, don’t max out.
- **Match velocity to site stage:**
  - **Stage 1, Sandbox** (keywords stuck beyond page 5; links produce no visible movement): goal = trust. Social profile links (all at once is fine, natural for a new business), ~30–40 business citations in a day or two (also natural), then **1–3 new-page links/week** (guest posts, podcasts, mentions, the web mostly links on new pages) spread across homepage/money/supporting pages until the sandbox exits (visible as a single-day rank jump + traffic surge).
  - **Stage 2, Trustworthy** (keywords on p1–4; links show effect in ~2–4 weeks): velocity must be *justified by traffic*, in the real web, every linker had to visit at least once. A 100-visitor/month site attracting 1,000 links/month looks engineered.
  - **Stage 3, Authority** (new content ranks p1–3 immediately): links still work but their individual effect drowns in noise; maintain steady acquisition.
- Big sites' velocity (e.g., The Spruce: ~1,500 referring domains/month) is only "natural" because traffic justifies it. Scale yours to your traffic reality.

## 2. Anchor text optimization, the per-page science

- Anchor text is **the** off-site lever you control after choosing where links come from.
- **Per-page, not site-wide:** Penguin evaluates anchor distribution per URL. A balanced homepage profile does nothing for an overcooked inner page.
- **Reverse-engineering process:**
  1. Pick the exact target keyword (intent matters, "best protein powder" ≠ "protein powder").
  2. Pull the anchor profiles of the top ~3 ranking pages.
  3. Classify every anchor: **target/exact-match, partial-match, brand, URL/naked, generic ("click here"), misc.**
  4. Average the top competitors → that’s the niche’s natural distribution → close the gap to it. (e.g., niche average 16% brand vs your 6% → build brand anchors next.)
- **Pitfalls:** ignore mega-authority outliers (a WebMD ranking with 2 links skews the ratios, exclude); keep one classification convention; don’t treat the distribution as universal, it’s niche-specific.
- **11 anchor rules (key ones):**
  - Never reuse the same target anchor, vary ("dog training" → "training for dogs" → "dog obedience tips"). Exact-match ≤2×, and only when volume is high enough that duplication is statistically plausible.
  - **SEO-title-tag anchors are the most natural** (forums/CMSs default to them, safe to reuse).
  - Match anchor type to link type (editorial context links get varied topical anchors; profiles/citations get brand/URL).
  - **Black-sheep effect:** if everyone on page 1 ranks with 3–5 links, being the one page with 20+ links flags you, you can’t out-link an authority deficit.
- **Link-gap analysis:** bucket competitor referring domains by authority tier (DR 20–30, 31–40, …), average across top competitors, subtract yours, multiply by a DR-scaling factor (exponential, a DR80 link ≫ DR20), exclude nofollow.
- **Tier-2 links** (links to your linking pages): use blended/varied anchors, goal = get the linking page itself ranking so it passes more value. (Note: tier-2 building is a manipulation tactic, evaluate risk.)

## 3. Topical authority (Koray Tugberk interview, the deepest take in the corpus)

- **Definition:** authority for specific entities & phrases built by answering *every relevant question* about a topic, count of pages doesn’t matter, **coverage of answers does**.
- His case studies: new domains, no links, no technical SEO, no titles/descriptions, still reached ~300k organic/mo in ~6 months purely via topical-map-complete publishing. Gains arrive gradually after ~3–4 months, then compound.
- **Why it works:** Google is a semantic/entity search engine, it extracts entities, attributes, and their possible questions, and can audit content accuracy. Covering an entity completely makes you the unique complete source → SERP refresh.
- **Building a topical map:** start from the entity type → its attributes → questions per attribute. Use PAA dynamically, PAA suggestions *shift based on what you click*, literally revealing Google’s internal topical map. Scrape competitor sitemaps.
- **Counterintuitive rule:** include zero-volume entities. If a node exists in Google’s entity graph and nobody fills it, filling it makes you a unique source. (But prune the map to the *correct context*, "apple" fruit vs company; use Google’s NLP API to check context classification.)
- **Internal linking = ontology:** link where two pages share an entity/attribute, the mutual point is the link point. (Ideal tooling: entity-aware interlinking, InLinks was closest at the time.)
- **Content briefs** specify per-question granularity scaled to competition; historical data (time covering the map) is itself a factor.
- Writers: law/philosophy/math/engineering backgrounds follow semantic briefs best.
- **Honest limit:** alone it may not beat strong-brand incumbents in brutal niches, pair with links/brand/historical trust.
- **Bottom-up publishing:** same lesson as Moz’s cluster guidance, bottom-funnel pages first (lower competition, faster revenue proof).

## 4. Other Diggity frameworks

- **SEO silos:** structure content in tight topical clusters, interlink only within/between related clusters, concentrates relevance signals.
- **QDF (Query Deserves Freshness):** some queries' SERPs reward recency (news, trending, seasonal, "best X 2025"). Detect: SERP dominated by recent dates → publish/update on a schedule matching the freshness window; otherwise evergreen depth wins.
- **Which pages to build links to:** prioritize money pages closest to revenue, then supporting pages that feed them via internal links; use internal links to distribute authority to pages too risky to direct-link.
- **Outreach conversion hack:** personalize the first line only (their content/site specifics); keep the ask template-rigid, biggest conversion lever is targeting (right prospects) > copy.
- **Case-study patterns across wins (3012%, +800%, +94% traffic):** topical-map content rollout → internal-link architecture → targeted authority links → refresh pages showing rank decay; beating HCU correlated with real experience signals (original imagery, named authors, first-hand testing).
- **Mistakes that get sites penalized:** over-optimized anchors, thin content at scale, link schemes, doorway-style programmatic pages, aggressive exact-match everything, neglecting technical debt.
- **Website audits (live):** recurring diagnoses, intent mismatch vs SERP, cannibalization, buried pages, weak supporting content around money pages, missing entity coverage vs leaders.


---


# Practitioners, Income School, Neil Patel, Brian Dean, Yoast

> Sources: `D:\seo-courses\transcripts\incomeschool-*`, `neilpatel-seo-2023`, `neilpatel-seo-unlocked`, `neilpatel-*` playlists, `briandean-*`, `yoast-*`.

---

## 1. Income School (Ricky & Jim), the no-link-building, content-first system

Contrarian philosophy: **backlinks are good; link building is a waste.** Their sites earned hundreds-to-thousands of referring domains organically (cookforfolks: 400+ links, zero effort; incomeschool.com: 34k+) by publishing hard-to-replicate content, original research, real experiments, own photos/data. Google’s link signals still matter, but the *method* is earning, not building. (Balance against Diggity/Ahrefs view: link building works in competitive niches; Income School niches are mostly low-competition hobby/info sites. For builder agents: allocate effort to linkable assets first, outreach second in weak niches; in competitive SERPs you will need deliberate acquisition.)

### Search analysis (their keyword-research replacement)
1. **Brand plan on a blackboard:** slice the niche into logical categories before touching tools (off-roading → vehicle types → buy/modify/maintain/use per type). This becomes the site’s topical structure.
2. **Pick max 3 subgroups to start**, depth in one lane beats scattered coverage; builds topical authority.
3. **Google Trends at topic level, 5-year view** to compare relative demand between subgroups, not keyword tools with dubious volumes.
4. Then find the question-style queries inside each subgroup (autocomplete, PAA, forums) and map them to post types.

### Post recipe taxonomy
- **Response posts (~1 hr to write, ~1,300 words):** directly answer one specific question, cheap "shots" to cover the grid fast.
- Winners get upgraded into longer staple/pillar posts only *after* proving themselves.

### Battleship method (content portfolio management)
- **Phase 1, Gridding:** methodically publish response posts across every area of the niche map (like calling coordinates in Battleship).
- **Phase 2, Analyze:** spreadsheet classifies each post by rank/pageviews: *hit* (keep + promote), *upgrade* (ranking but improvable, fix snippet/content), *miss* (flop, never touch again; one audit found ~100 misses), *maintainer* (ranking, monitor only).
- **Phase 3, Invest in hits:** a response post stuck at #4 that’s proven demand → now an 8-hour pillar post is justified ("it’ll bring 1,000 visits/mo for years"). Never spend pillar effort on unproven topics.

### Answer targets, featured-snippet engineering (their signature tactic)
- ~76% of searches show a featured snippet; everyone writes the post, almost nobody engineers the snippet.
- For each post write a 2–3 sentence **answer target**: a standalone, encyclopedia-toned (semi-formal, no slang/opinion/personality, "add flavor? make it vanilla") direct answer, starting with the noun not a pronoun (no "it/they" openers, must make sense without the question).
- **Tee it up:** the preceding sentence signals the answer is coming (context, "here’s what…"); don’t restate the question verbatim (old tactic, FAQ-style formatting made it less necessary).
- **Bold it:** standout formatting helps readers AND signals importance to Google.
- In their testing, every snippet won was the intended answer target, Google used exactly the engineered sentence.
- 4 snippet types: paragraph (~50%), list, video, table (underserved, deliberately build table snippets for comparison queries).

---

## 2. Neil Patel, SEO Unlocked (7-week program) + SEO 2023

Structured 7-module curriculum: SEO basics → keyword research (Ubersuggest workflow) → on-page → content marketing → link building → branding → advanced/CRO. Distinct contributions beyond standard on-page advice:

- **Content framework:** idea-generation spreadsheet + "20/20 rule book"; content body uses a **trigger section**, problem → dive deep → solution (problem-solution structure converts).
- **Branding as SEO:** brand = collective perception built from 4 elements, what it says about itself, what it does, what others say, how others interact. Consistency across every touchpoint (support, sales, email, video) → brand searches → ranking moat (branded query volume is itself a demand signal Google rewards).
- Spend effort on promotion, not just creation, publishing without distribution is wasted.
- Worksheets-driven approach: USP, value proposition, offer, then inject those insights into title tags, copy, design.

## 3. Brian Dean (Backlinko), named link-building systems

- **Skyscraper technique:** (1) find content already ranking for your keyword, (2) create something significantly better, longer, better designed, more current, more thorough ("deserves to rank #1"), (3) promote specifically to sites that already link to the inferior piece, they’ve demonstrated link-willingness on the exact topic. Cited results: +110% traffic/14 days; reader Emil +59%, #1, 10k PV/mo.
- **Guestographics:** (1) publish a quality infographic, (2) Google the topic → sites already writing about it, (3) two-touch outreach, first ask *"want to see it?"* (no begging; flip the script), (4) offer a **unique custom introduction** as the bribe, the #1 barrier to sharing an infographic is the site needing unique intro text, so remove it, (5) contextual links land. (+175% organic; Perrin +963% in 6 weeks, 1,000+ referral visits.)
- **Moving Man Method:** find companies that rebranded/moved/shut down → their old URLs still return 200 (dead content, live page) so broken-link checkers miss them → export their backlink profile → notify linkers ("your link is dead") + pitch your resource as replacement. His BlueGlass campaign → .edu + PR5 resource-page links at high conversion.
- **Broken link building (classic):** find dead links on relevant resource pages → notify + pitch replacement, helping them fix their site is the conversion lever.
- Core pattern across all: **targeting > copy**, contact people who demonstrably link to your topic; make the ask a favor to them, not to you.

## 4. Yoast, WordPress-era fundamentals that still bite

- **Categories vs tags:** every tag auto-creates an archive page; thin 1-2 post archives hurt SEO. Few broad categories; tags only for genuine cross-cutting themes; noindex thin archives.
- **Duplicate content:** same content on multiple URLs forces Google to cluster and pick a canonical, splitting your own signals. Consolidate/canonicalize (don’t duplicate product/category descriptions across URLs).
- **Headings:** strict hierarchy, one H1, H2 sections, H3 subsections, for scannability, screen readers, and Google’s structure parsing.
- **Internal links:** discovery + importance signals; every important page reachable in ≤3 clicks (aligns with Google guidance).
- **Transition words** (because, however, for instance): readability ↑ → engagement signals ↑.
- **Rebuild decision:** 10-year-old static HTML site → move to WordPress/CMS only if you’ll actually keep updating, a stale rebuilt site is no better.
- **Seasonal playbook (Black Friday):** prep metadata weeks ahead on top-10 money pages, swap seasonal copy in/out around the event, small effort, repeatable wins.
- **AI-era items (2025 explainers):** brand visibility in ChatGPT/Gemini answers is measurable (Yoast AI Brand Insights: generate the queries people ask AI → track presence vs competitors); NLWeb (Microsoft open project), structured data + APIs making sites conversational-agent-readable; social profiles/posts are an off-page relevance surface.


---


# Mega-Course Distillations, AI SEO, Programmatic, Ecommerce, Technical, Link Building, Keyword Research, Local

> Sources: `D:\seo-courses\transcripts\mega-*`, 47 long-form single-video courses. Extracted the unique frameworks; generic beginner content deduped against earlier sections.

---

## 1. AI search: which page types actually win (Ross Hudgens / Siege Media, 96-min deep dive)

- **Versus/comparison articles are the page type driving the most AI-search visibility.** LLMs run "comparison work" inside down-funnel prompts ("me vs competitor") that never show in tracked prompts, you get 100% visibility in the answer. Clients resist publishing them ("don’t talk smack"); use the data to convince them.
- **Vs-page anatomy:** "us vs them" H1 + context + CTA above the fold → **comparison table very high on the page** (feature checkmarks, tractable for users, LLMs ingest tables cleanly) → deeper detail below. Base comparisons on what’s actually coming up in sales calls, not just search volume.
- **Triopoly comparisons:** in 3-player markets, people genuinely search "A vs B vs C", and if you’re the 4th player, a 3-way article is how you enter the conversation.
- **Hub pages:** HubSpot links a "software comparisons" library in the footer, quiet IA move feeding equity to all transactional content. But **do not template-vs every competitor**, garbage templated comparison pages won’t convert and can trigger doorway-style problems (see §3).
- **Most defensible content vs AI Overviews:** complex interactive assets, calculators, tools, configurators. Simple calculators get replicated inside ChatGPT/Claude; **complex multi-input calculators simply aren’t getting AIO’d**. (Same conclusion as the ContactOut pSEO data: value-adding utility pages survive.)
- **Citation metrics, used correctly:** don’t weight one LLM’s citations equally, a URL driving thousands of Perplexity citations may be <1% of real influence. Cross-reference citation share with actual traffic/traffic value to filter "AI slop" citation sources (high citations, zero real traffic, won’t last). **Retrieval is good; citation is better.** AI-slop cites are a trap metric.
- **Reddit still matters** for AI visibility, prominently cited; they’re building their own LLM surface.
- KOB analysis (Keyword-Opportunity-to-Benefit) framework for prioritizing which comparison/product pages to build first.

## 2. Programmatic SEO, scale without the doorway penalty

- **Model:** keyword = head term + modifier ("{person} email", "{city} vacation rental", "{word count} pages"). Verify head-term volume (Ahrefs, not just Google KW planner ranges) and confirm modifiers add up, pSEO pages individually carry little authority, so target low-KD head terms.
- **THE LINE, doorway abuse (Google’s own definition):** pages created to rank for specific/similar queries that lead users to intermediate pages less useful than the destination; multiple pages differing only by a keyword/location swap; pages existing as islands outside normal navigation. Real-world hit: HVAC company, hundreds of suburb pages with only the place-name swapped → March 2024 core update → 80% of doorway pages lost rankings, **–63% organic traffic in 30 days**. Recovery only after consolidating into comprehensive, genuinely location-specific pages with unique content + per-area reviews.
- **Doorway test (builder agents, apply before generating):** Is the page solely designed to rank for a keyword/location? Does it offer little unique content? Do multiple pages differ only by one variable? Are the pages hidden from main navigation? Would you be embarrassed to show them to real visitors? → Any "yes" = doorway risk.
- **pSEO done right:** the variable must have *dependencies*, content must meaningfully change per value (LA HVAC page discusses LA’s climate, served neighborhoods, local reviews, not a place-name find/replace). bookscal.com example: each template page is a working calculator + related interactive suggestions → ranks for 629 keywords, 81 in positions 1-3. **Utility per page, not text per page.**
- Surface pSEO pages through a visible hub page in nav, never orphan islands.
- Write natural phrases: "HVAC repair in Los Angeles" not fragment-H1 "HVAC repair Los Angeles", fragment stuffing reads as spam to users and Google.
- **Data sourcing for real pSEO:** buy pre-scraped datasets (Bright Data, Coresignal, LinkedIn/Crunchbase/Glassdoor/Yelp, $k–$100k), public data (census, SEC, court records, WHOIS), or scrape transcripts/podcasts then **transform** (ContactOut fed LinkedIn scrapes through an LLM to write unique articles; derived company tech-stacks from job postings). Scraped republishing without transformation = thin content that won’t index/rank.
- **Indexing at scale is the real bottleneck:** indexation rate tracks domain authority/age, new domains get a fraction indexed. Drive it with internal links + sitemaps; only keep indexable pages that can earn traffic; monitor per-template conversion ("Katy Perry email" pages pulled traffic, ~zero B2B conversions).
- Link building for pSEO: homepage-level authority (digital PR, reviews-for-links, selective 3-way exchanges) lifts the whole page fleet.

## 3. Ecommerce SEO, revenue-first playbook ("$40M advice")

- **Only 4 technical things matter:** crawlable + properly indexed; no broken links; good mobile+desktop UX; clear hierarchy. Meta descriptions: not a ranking factor, they move CTR only (optimize for clicks, don’t obsess). Everything else is agency noise.
- **Money pages > blog traffic.** Ranking #1 for "why use natural face wash" brings traffic that converts ~0%. Case: puzzle brand paid two agencies $120k over 2 years for blog content → never cleared $2k/mo; building **40 new collection pages** broke $10k/mo in ~4 months. KPI = organic search revenue, top-3 positions for transactional keywords, not traffic graphs.
- **Collection-page expansion is the ecommerce growth lever:** map every real modifier customers search (color/size/material/style/occasion/audience combos) → unique collection pages with genuine merchandising differences.
- **Links:** 10–20 right backlinks beat 1,000 garbage links. **Link insertions on pages that already have traffic** > guest posts (a guest post has no traffic guarantee; a live ranking page does, and insertions run ~20-30% cheaper, ~$100–400). Marketplace markups add 20-60%, direct outreach when cash-constrained.
- Realistic take on link earning: "content so good it earns links" is a DR70-80 privilege; everyone else needs digital PR (~$5k/mo) or paid placements, budget honestly.
- **Tool stack that was enough for $40M:** Semrush (all-in-one; competitor links/copycat), Google Keyword Planner (free, CPC ranges double as a competition proxy: high CPC ≈ profitable ≈ hard organic), Screaming Frog (free to 500 URLs; status codes, titles, H1s, canonicals, the one technical tool that matters).
- **80/20:** money-page build/optimization + quality links = ~20% of effort, drives the lion’s share of revenue. Reporting/tech minutiae/blog posts/internal-link busywork = the other 80% of time, keep it contained.

## 4. Technical SEO, Rank Math’s 9 practices + live audit patterns

- **Orphan pages:** find them (Rank Math’s internal-links-in icon; Screaming Frog Orphan report) → every index-worthy page needs ≥1 internal link.
- **Content-structure models (pick deliberately):** silo (strict topic compartments, max topical clarity), hub-and-spoke (hub links spokes, spokes link back), content library (topic repositories), topic gateway (category overview → subtopics), database model, glossary model. Map yours in Sheets/Miro before scaling, **prevents cannibalization** (Rank Math flags duplicate focus keywords; fix by updating the existing page, not adding another).
- **404 pages:** Quality Rater Guidelines classify a plain 404 as medium quality; a 404 with search + helpful links can rate high/highest. Add search + popular-links to every 404.
- WordPress gotchas: theme switches wipe footer/social links (re-add after every theme change); "Discourage search engines" checkbox in Reading Settings kills the whole site; a canonical set under Advanced can silently mark a page "duplicate of…".
- **Indexation debugging order:** GSC Pages tab → find should-be-indexed page → URL Inspection (refreshes faster than the Pages report) → was it crawled? (robots block? removal tool? server error?) → if crawled-not-indexed: noindex? canonical pointing elsewhere? → fix → "Validate fix" + Request Indexing (or IndexNow/Indexing API via plugin) → screenshot the live-URL test to see what Google actually renders.
- **Duplicate content:** SiteLiner, >20% match on indexable money pages needs action (whole-paragraph duplication); ignore shared single words/category-listing overlap; fix legit dupes (print versions, product variations) with canonicals, fix thin/poor dupes with 301s.
- **XML sitemaps:** always have one, split by content type under a sitemap index, lastmod dates must be truthful (don’t fake freshness, Google corroborates).
- **Schema’s new job:** beyond rich snippets, it’s how AI Overviews/answer engines parse your content. Match schema type to page purpose (Article, Book, Course, Product, Review+pros/cons); fill it completely.
- Mobile: ~60% of traffic, mobile-first indexing, test with DevTools responsive toggle + Bing’s mobile-friendliness tool (flags tap-target spacing etc.); a theme from "donkey years ago" = revamp or replace.

## 5. Link building, Nathan Gotch’s 10-year distillation

1. **Build a personal brand around a real expert**, expert = better than the public; *authority* = expert with influence. (Gary Vee: 16k RDs to his personal site, 2.6k flowing to his agency.) The person must be willing to market themselves.
2. **Blue-ocean linkbait:** steal *angles* proven in other industries before competitors copy, custom design, epic guides, original data, free tools/software (ChatGPT builds them fast, his script-timer took 1 hour), weird content, rankings/list posts, deep technical pieces. Untapped angle in your niche ≈ 100% of the links; copied angle = scraps.
3. **Relevancy pyramid:** relevance outranks DA/DR/traffic metrics (echoes John Mueller: one relevant link can outweigh millions of junk ones). Expand prospect pools via Ahrefs traffic-share-by-domains, `site:` operator topic checks, adjacent subreddits, then ChatGPT to rank tiers.
4. **Merger technique (advanced/greyer):** acquire undervalued sites with strong editorial link profiles → 301 each URL by link distribution (homepage-heavy → a dedicated inheritor page; distributed links → matching target pages) → internal-link from the inheritor. Price the domain by summing its backlink values first. Works, but it’s acquisition-and-redirect manipulation; weigh against Google’s spam policies.
5. Competitor backlink gap (Semrush Backlink Gap / Ahrefs Link Intersect), domains linking to rivals but not you.
6. Similar-content outreach: only works with a genuinely unique angle.
7. Resource pages via `"keyword" + "resources"|"useful links"|intitle:links` search strings.
8. **Digital PR** (journalist pitching à la Search Intelligence/HARO-style): homepage links are underrated, equity flows to all pages if IA is clean.
9. Internal-link leverage: identify your most-linked pages (Semrush Indexed Pages / Ahrefs Best by Links) → add contextual internal links from them to money pages.
- Bonus trust signals: reviews velocity + technical soundness make links "stick" (Google trusts the target).

## 6. Keyword research, Ahrefs 3-step + cross-source consensus

- **Step 1, Gather:** seed keywords (ask AI for 20 expansions) → Matching Terms; your own Organic Keywords → dropped rankings → related questions + "also rank for"/"also talk about" terms to rebuild topical coverage; **Content Gap** (competitors' top-10 where you’re absent from top-100, sort by traffic); **paid keywords**, competitors paying CPC have validated profitability (sort by CPC; high paid+organic presence = proven converter).
- **Step 2, Filter by metrics:** intent → funnel stage (top/mid-funnel informational+commercial for new sites; transactional for money pages); **"identify intents"** shows the traffic split across intents, one page can serve multiple; SV≥500 + KD≤10 preset = low-hanging fruit for new sites; SERP-check that you can actually win (domain mix, SERP features).
- **Step 3, Prioritize:** traffic potential (parent topic’s total traffic, not the single keyword volume, one page ranks for thousands of variants); business value score; effort. This matches Income School (3 subgroups, questions-first), Moz (striking distance + cluster), and HubSpot (money keywords first for AI audit).
- Cross-source consensus: **CPC as competition/value proxy** (also the $40M ecommerce method); question-based queries are the future (PAA, AI prompts); keyword tools give ranges, treat volumes as directional, not gospel.

## 7. Local SEO, synthesis across the mega local courses + Ahrefs

- **GBP/Maps ranking trio (consistent across all courses):** relevance (categories + services + keyword-complete profile), distance (can’t optimize, but service-area honesty), prominence (reviews + citations + links + on-site signals).
- Review velocity beats review count: steady cadence, respond to all, keyword-rich service mentions in responses; never gate or buy reviews.
- NAP consistency + citations as table stakes (audit once, maintain); local schema (LocalBusiness + sameAs); location pages need *real* per-location uniqueness, the HVAC doorway case above is the local-SEO trap.
- "Local SEO systems" angle (33-min Claude-assisted video): automate review-request flows, GBP posting cadence, citation auditing, systemization > heroics.
- City ranking fast-track (mega-ranking-2026): proximity matters but relevance/prominence win, service pages per service (not per city unless genuinely differentiated), embed GBP, local links/sponsorships, review pipeline.

## 8. Ranking-in-2026 quick hits + the AART system

- "How I’d rank in the AI era" consensus: brand + entity clarity (schema, about/author pages, consistent mentions), content designed to be *cited* (answer blocks, stats, quotes), presence on platforms LLMs cite (Reddit/YouTube/industry pubs), technical baseline fast + crawlable.
- AART system (160-min course): Audit → Assets → Rank → Track, cyclical process framing; nothing contradictory with the playbooks above.
- Google Digital Marketing certificate module: solid beginner canonicals, SEO = making pages useful to people, on-page/off-page/technical trio, no guaranteed rankings.


---



# APPENDIX A, FULL COURSE TEXT: Ahrefs AEO (the example file, merged in)

# Ahrefs Academy, Answer Engine Optimization (AEO) Course
**Full transcript / course text**, by Sam Oh (VP of Marketing, Ahrefs)
Source: https://ahrefs.com/academy/aeo-course (12 lessons, ~1h 26m)

## About this course
Teaches how AI search works and how to get your brand recommended across every major AI platform
(ChatGPT, Google AI Overviews, Perplexity, Gemini).

### What you’ll learn
- The fundamentals of how AI search works and why AEO matters
- How different AI platforms decide what to cite
- How to find gaps between your brand and competitors in AI search
- How to do keyword and prompt research for AEO
- How to create content that AI wants to cite
- How to earn brand mentions across the web
- How to optimize YouTube videos for AI visibility
- Technical AEO best practices
- How to track AI traffic and measure your progress

### Course structure
- Intro: What is AEO? (6:34)
- Module 1, How AI Search Works: lessons 1-1 to 1-3 (~21 min)
- Module 2, AEO Strategy: lessons 2-1 to 2-2 (~16 min)
- Module 3, AEO Execution: lessons 3-1 to 3-4 (~27 min)
- Module 4, Measuring AEO: lessons 4-1 to 4-2 (~15 min)

*Note: Lessons 0-1 through 2-2 are the official written lesson text from the course pages.
Lessons 3-1 through 4-2 are transcripts of the video audio (those pages have no written version).*


---

# Introduction


## Lesson 0-1: What is AEO? (Course Intro)

Before learning how to optimize your brand’s visibility for AI search, it’s important to learn why
AEO matters and how it’s different than SEO.

## Why AEO matters right now?

As of [December 2025](https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/), the presence of an AI Overview on Google reduces the click-through rate
(CTR) for the number one ranking page by 58%. That means for every 100 clicks a page used to get,
Google now keeps 58 of them.

Some other numbers to consider:
* ChatGPT alone has 900 million weekly users
* It handles roughly 12% of Google search volume
* AI traffic to websites has grown 9.7 times in the past year

You can either ignore this shift and watch your traffic shrink, or learn how it works and capitalize
on it.

Here’s what most people miss: in [June 2025](https://ahrefs.com/blog/ai-search-traffic-conversions-ahrefs/), AI search accounted for just 0.5% of Ahrefs'
traffic, but it drove 12.1% of sign-ups. That’s a 23 times higher conversion rate than organic
search. Meaning AI search traffic does convert better.

## How AEO differs from traditional SEO

In traditional SEO, you optimize web pages to rank higher in a list of search results. You’re
competing for positions. The user sees your blue link, clicks it, and lands on your site.

In AEO, there is no list. The AI reads from dozens of sources, synthesizes an answer, and decides
who to mention or cite. You’re not competing for a position, you’re competing for a mention.

The key thing to understand is this: AEO doesn’t replace SEO, it builds on top of it. Think of SEO
as your foundation and AEO as future-proofing your online presence.

## Is AI going to kill SEO?

This is not the end of SEO. [Ethan Smith from Graphite talked on Ahrefs podscast ](https://www.youtube.com/watch?v=ejylKNKAQ4s)about the
zero-sum bias, a cognitive bias based on the assumption that if something new goes up, something old
must go down.

*(screenshot)*

The same thing happened in 2010 when the App Store blew up. Everyone said the web was going away,
that mobile apps were taking over, that only old people used websites. Mobile apps did blow up, that
promise was real, but the web didn’t die. It grew alongside these apps.

AI search is likely going to follow the same story. It is growing fast, and the numbers shown
earlier are real, but Google still processes billions of searches per day and traditional organic
traffic still drives massive business value.

The smart move isn’t to panic, it’s to play both games.

## Most businesses haven’t started yet

Ahrefs looked at its own data and found that its content and product pages have been mentioned
[7,470 times across 2,309 pages in AI search](https://ahrefs.com/blog/geo-generative-engine-optimization/), without any effort to optimize for it. That’s
happening organically because Ahrefs has been doing solid SEO and creating content that people find
useful.

[Screenshot of the Ahrefs Site Explorer overview for ahrefs.com. The "AI citations" section is
highlighted, showing the number of times the site has been cited by AI assistants like Google AI
Overview, ChatGPT, and Perplexity.]

Now imagine what happens when a business starts intentionally optimizing for this: knowing which
platforms cite what, understanding how AI decides who to recommend, and actively building the
signals that make AI want to mention a brand.

That’s exactly what this course teaches.

## Course structure

The course is broken down into four modules plus this introduction.

1. Module #1: How AI search actually works. Covers the mechanics of how AI finds content, how it
decides what to cite, how different platforms like ChatGPT, Google AI Overviews, and Perplexity
differ from each other, and what AI visibility looks like for different types of businesses. This is
the foundation that makes everything else make sense.

2. Module #2: Strategy. Covers the data behind why brand mentions are the single strongest lever for
AI visibility, stronger than backlinks, domain rating, or anything else. Also covers keyword
research for AEO, prompt research, and how to figure out where a brand stands versus competitors.

3. Module #3: Execution. This is where the actual work happens: how to create content that gets
cited, how to earn mentions and citations from other sites, YouTube optimization for AI visibility,
and the technical work needed to make sure AI can actually find and crawl content.

4. Module #4: Measurement. Covers how to set up analytics for AI traffic, how to run competitive
audits to find brand gaps, and how to evaluate whether AEO is actually worth the investment for a
given business.


### Further reading

Check out some of the articles on the [Ahrefs blog](https://ahrefs.com/blog/category/ai-search/) to learn about AEO:
* [Guide to Answer Engine Optimization](https://ahrefs.com/aeo)
* [Answer Engine Optimization: How to Win in AI-Powered Search](https://ahrefs.com/blog/answer-engine-optimization/)
* [Generative Engine Optimization: Growth Strategies and Metrics For the AI Era](https://ahrefs.com/blog/geo-generative-engine-optimization/)
* [Does AI Search Traffic Convert Better Than Traditional Search? For Ahrefs, Yes: 0.5% of Visitors
  Drove 12.1% of Signups](https://ahrefs.com/blog/ai-search-traffic-conversions-ahrefs/)
* [Update: AI Overviews Reduce Clicks by 58%](https://ahrefs.com/blog/ai-overviews-reduce-clicks-update/)
* [Top SEO myths debunked with real data | Ethan Smith (Graphite)](https://www.youtube.com/watch?v=ejylKNKAQ)

*(Source: https://ahrefs.com/academy/aeo-course/lesson-0-1)*



---

# Module 1. How AI Search Works


## Lesson 1-1: How AI Search Engines Work

What happens between the moment someone types a prompt into an AI search engine and the moment that
engine cites (or ignores) your brand?

Understanding this process matters because every answer engine optimization tactic only makes sense
once you understand the mechanics behind AI search.

## Where AI gets its information

AI search engines pull from two very different sources of information:
1. Training data (static) is the massive collection of text an AI model was originally trained on:
   books, websites, PDFs, social media, YouTube transcripts. It’s essentially a snapshot of the
   internet at a point in time.The problem with training data is that it’s static. It gets updated
   only every six months or so.
2. Real-time retrieval (live web searches) powered by RAG (retrieval-augmented generation). When
   ChatGPT or Google’s AI Mode needs fresh information, or when a question is too specific for
   training data alone, the system searches the web using APIs, pulls back a set of pages, reads
   through them, and generates a response based on what it found.

*(screenshot)*

This matters because it means there are two distinct ways to influence what AI says about your
brand:
* Be mentioned so widely across the web that you become baked into the training data itself
* Make sure your content shows up when the AI searches the web in real time

That second path is SEO. Ranking in Google, earning backlinks, and creating quality content directly
influence whether AI picks up your pages during real-time retrieval.

## Query fanout: one search becomes many

AI doesn’t just search for the exact phrase you typed in.

Search engines have evolved through a few stages:
* One-to-one: one query, one set of results
* Many-to-one: different queries, like "Sydney plumber" and "plumbing service in Sydney," return the
  same results.
* One-to-many: AI search’s current model, where a single query is expanded into many; this is called
  query fanout

*(screenshot)*

For example, a prompt like "plan me a 5-day trip to Japan in November" gets fanned out into dozens
of smaller, long-tail subqueries running simultaneously behind the scenes, such as: "best
neighborhoods to stay in Tokyo", "November weather in Kyoto", "Japan Rail Pass worth it".

The AI pulls information from multiple sources across the web for each of these subqueries and
combines everything into one complete answer.

*(screenshot)*

If your content ranks for those niche, specific queries, your brand has a much better chance of
being included in the AI’s final response.

This is a major shift from traditional SEO, where you could optimize one page for one target keyword
and call it a day. In AI search, you need to be relevant across an entire topic, arguably across an
entire niche.

## Seeing fanout queries yourself

You can actually see these fanout queries under the AI responses report in[Ahrefs' Brand Radar](https://ahrefs.com/brand-radar/)
for ChatGPT and Perplexity prompts.

*(screenshot)*
Tip

Fanout queries are not like traditional long-tail keywords. They’re synthetic, generated by the AI
in the moment, and inconsistent, meaning the same prompt can trigger different fanouts every time.
Over 95% of them have zero search volume because real humans would never type them.

So don’t treat fanout queries as a new keyword list to optimize for, but as a window into what
topics the AI considers.

## How AI decides who to cite

In traditional search, rankings are relatively stable. If you’re number three for a keyword today,
you’re probably going to be around there tomorrow. AI citations don’t work that way, they’re
probabilistic.

Patrick Stox explained it well in [Ahrefs Podcast](https://www.youtube.com/watch?v=bHDrtW75Wlo): AI outputs are built on probabilities on top
of probabilities on top of probabilities.

In practice, this means if you ask the same question five times, you might get cited three out of
five times. There’s no fixed position to rank for. This is why the industry talks about AI
visibility rather than AI rankings. It’s more like a probability distribution than a leaderboard.

That said, patterns do emerge in what gets cited more often, based on data Ahrefs has studied:
* Consensus matters: if multiple sources on the web say the same thing about your brand, AI is more
  likely to repeat it. The more places your brand is mentioned consistently, the higher the
  probability AI picks it up.
* Freshness matters: [AI-cited content tends to be about 25% fresher](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content/) than what typically shows
  up in a traditional SERP. The AI actively looks for recent information, especially for topics that
  change.
* Authority still matters: pages that rank well in traditional search have a major head start. 38%
  of AI Overview citations come from pages already in the top 10 of Google.


### Further reading

Check out some of the articles on the [Ahrefs blog](https://ahrefs.com/blog/category/ai-search/) to learn about AEO:
* [Guide to Answer Engine Optimization](https://ahrefs.com/aeo)
* [Answer Engine Optimization: How to Win in AI-Powered Search](https://ahrefs.com/blog/answer-engine-optimization/)
* [Generative Engine Optimization: Growth Strategies and Metrics For the AI Era](https://ahrefs.com/blog/geo-generative-engine-optimization/)
* [What is Query Fan-Out? Understanding the Hidden Queries Driving AI Search](https://ahrefs.com/blog/query-fan-out/)
* [Update: 38% of AI Overview Citations Pull From The Top 10](https://ahrefs.com/blog/ai-overview-citations-top-10/)
* [New Study: AI Assistants Prefer to Cite “Fresher” Content (17 Million Citations Analyzed)](https://ahrefs.com/blog/do-ai-assistants-prefer-to-cite-fresh-content/)
* [World’s Top SEO Expert Busts AI Search Myths | Patrick Stox (Ahrefs)](https://www.youtube.com/watch?v=bHDrtW75Wlo)

*(Source: https://ahrefs.com/academy/aeo-course/lesson-1-1)*


## Lesson 1-2: How Ranking in Google AI Overviews, ChatGPT, and Perplexity are Different

AI search platforms don’t find, rank, or cite content the same way. Treating "AI search" as a single
target and optimizing the same way across the board means leaving visibility on the table.

## How different are these platforms, really?

[An Ahrefs study found a 14% overlap](https://ahrefs.com/blog/top-mentioned-sources-are-not-shared-across-ai-assistants/) for the top 50 most cited domains across Google AI
Overviews, ChatGPT, and Perplexity. Out of those 50 domains, only seven appeared on all three
platforms.

[top 50 mentioned brands overlap count by ai assistants]

If you’re only optimizing for one platform, you could be effectively invisible on the others.

### Google AI Overviews

Google AI Overviews favor authoritative, established sites (think health, finance, and encyclopedic
content), plus Google-owned properties like YouTube.
* YouTube accounts for about 5.6% of all AI Overview citations
* Reddit is also heavily featured

[Ahrefs Study showing YouTube mention and YouTube mention impressions correlations with AI
Visibility in ChatGPT, AI Overviews, and AI Mode]

### ChatGPT

ChatGPT leans heavily toward publishers and media. Some of the most cited domains include: Reddit,
Wikipedia, Amazon, Forbes, Business Insider, and Wired.

[The median domain rating of ChatGPT’s top cited pages is 90](https://ahrefs.com/blog/chatgpts-most-cited-pages/#ranking-cited-pages), meaning it pulls overwhelmingly
from high-authority publishers. This is partly due to licensing deals OpenAI has with certain
outlets.

### Perplexity

Perplexity is the most aligned with traditional Google search results.

About [28.6% of Perplexity citations come from pages that rank in Google’s top 10](https://ahrefs.com/blog/ai-search-overlap/). Compare that
to ChatGPT, which overlaps with Google’s top 10 only about 8 to 10% of the time.

[Ahrefs research using Brand Radar: Only 11% of AI citations overlap with Google’s and Bing’s top
10]

If you’re already ranking well in Google, Perplexity is likely where you’ll see the most immediate
AI visibility.

### Google AI Mode

AI Mode and AI Overviews are both Google products, but they cite different sources:
* The citation overlap between AI Overviews and AI Mode is only 13.7%, despite their answers being
  86% semantically similar. Translation: they say similar things, but pull from completely different
  sources.
  
  [AI Overviews vs AI Mode response similarity indicating 16% word similarity, 13.7% citation
  similarity and 86% semantic similarity between the two.]
* AI Mode’s top cited domain is YouTube by a wide margin, followed by Google and Wikipedia.
* AI Mode cites Quora 3.5 times more than AI Overviews does, and it pulls much more heavily from
  social platforms like Facebook and Instagram.
  
  [AI Overviews vs AI Mode citation overlap according to Ahrefs' data study]

The takeaway: AI search is not one thing. Each platform has its own index, its own biases, and its
own preferences for the kinds of sources it cites.

## Which platform should you focus on?

Rather than trying to optimize for everything at once, it helps to be strategic. Two things to
consider:

### 1. Market share

Google’s AI features and ChatGPT currently account for the vast majority of AI search traffic.
Perplexity is growing but still represents a fraction of the volume.

If you have to prioritize: focus on Google and ChatGPT (at least at the time of recording this
course).

### 2. Overlap with your existing SEO

Consider how much overlap exists between what you’re already doing in SEO and what each platform
rewards:
* If you’re already ranking well in Google, you have a natural head start with AI Overviews and
  Perplexity.
* AI Overviews are increasingly pulling from pages outside the top 10, including YouTube and Reddit,
  meaning the connection between Google rankings and AI citations is weakening.
* ChatGPT is much more interested in publisher authority and editorial content. If your brand gets
  mentioned on Forbes, in a Reddit thread, or on a niche review site, that may matter more for
  ChatGPT visibility than your own page’s Google ranking.

Tip

Many of the top cited domains in AI search get no traditional search traffic at all. AI visibility
is its own game.

## Checking which domains are cited for your brand

You can see the top domains for a specific platform using Brand Radar:
1. Run a blank search in [Brand Radar](https://ahrefs.com/brand-radar/).
2. Go to the Cited Pages report > All domains.
3. Filter using your desired AI platform.

*(screenshot)*

Typically, the websites citing your brand differ across each platform. A domain that shows up
heavily in AI Overviews might barely appear in ChatGPT, and vice versa. This is why a
one-size-fits-all approach to AEO doesn’t work.

## Putting this into practice

You don’t need completely separate strategies for each platform. Fundamentals like creating quality
content, earning mentions, and building topical authority help across the board.

But knowing where each platform pulls from helps you prioritize your efforts. For example:
* If ChatGPT is a big traffic driver for your niche, focus on getting mentioned on high DR editorial
  sites and publishers.
* If AI Overviews matter more, YouTube and Reddit should be on your radar.

Specific strategies for these platforms are covered in modules two and three. For now, the key point
is that each platform is its own ecosystem with its own rules.


### Further reading

Check out some of the articles on the [Ahrefs blog](https://ahrefs.com/blog/category/ai-search/) to learn about AEO:
* [Guide to Answer Engine Optimization](https://ahrefs.com/aeo)
* [Answer Engine Optimization: How to Win in AI-Powered Search](https://ahrefs.com/blog/answer-engine-optimization/)
* [Generative Engine Optimization: Growth Strategies and Metrics For the AI Era](https://ahrefs.com/blog/geo-generative-engine-optimization/)
* [86% of Top Mentioned Sources Are Not Shared Across ChatGPT, Perplexity, and AI Overviews](https://ahrefs.com/blog/top-mentioned-sources-are-not-shared-across-ai-assistants/)
* [Update: 38% of AI Overview Citations Pull From The Top 10](https://ahrefs.com/blog/ai-overview-citations-top-10/)
* [67% of ChatGPT’s Top 1,000 Citations Are Off-Limits to Marketers (+ More Findings)](https://ahrefs.com/blog/chatgpts-most-cited-pages/)
* [Only 12% of AI Cited URLs Rank in Google’s Top 10 for the Original Prompt](https://ahrefs.com/blog/ai-search-overlap/)
* [Are AI Mode and AI Overviews Just Different Versions of the Same Answer? (730K Responses
  Studied)](https://ahrefs.com/blog/ai-overviews-vs-ai-mode/)

*(Source: https://ahrefs.com/academy/aeo-course/lesson-1-2)*


## Lesson 1-3: What Is AI Visibility? The 3 Types Every Marketer Needs to Know

AI visibility gets thrown around like it means one simple thing: getting a link in an AI response.

But that’s only part of the picture, and it might not even be the most valuable part for your
business.

## The three types of AI visibility

As you go through these, think about which one applies to your business.

1. Cited and linked: This is the one everyone thinks about. AI includes a link to your page in its
sources, and the user can click through and land on your site. This is the best case for direct
traffic, and it’s the easiest to measure.

2. Mentioned, but not linked: AI says your brand in its response, but doesn’t give the user a link
to click. You don’t necessarily get direct traffic, but the user now knows your name, and if they’re
interested, they’ll go search for you. This is basically word-of-mouth recommendation at scale.

3. Not visible at all: AI doesn’t mention you and doesn’t link to you. You’re simply not in the
conversation. This matters to know because you can’t fix what you don’t know is broken.

### How often mentions actually include a link

Most people assume that when AI mentions a brand, it’s probably including a link. That’s not the
case.

On average, only about [28% of AI mentions include a link](https://ahrefs.com/blog/ai-citations-vs-impressions-study/), meaning 7 out of 10 times your brand
comes up in an AI response, there’s no link attached.

This varies a lot by platform:
* Perplexity: about 51.6% of mentions get a link
* AI mode: around 36.8%
* ChatGPT: about 26.9%
* AI overviews: only 10.7% (9 our 10 times there’s no link)

[Bar graph: Mentions of "Ahrefs" in AI indexes. Orange bars: mentions without Ahrefs cited. Pink:
with Ahrefs cited. Perplexity cites Ahrefs most.]

## Linked mentions skew toward higher-traffic queries

You might assume this means most AI visibility is low value since there’s no click.

But when mentions are weighted by search volume (similarly to how people ask similar questions), the
mentions that do include a link tend to appear on much higher traffic queries.

For example:
* On Perplexity, links show up in about 78% of total impressions, even though they’re only in 51% of
  individual mentions.
* On Gemini, it’s even more dramatic: links appear in 71% of impressions, despite being in only
  16.8% of mentions.

[Bar graph comparing Ahrefs citations in AI tools. Orange: not cited, pink: cited. Higher citation
in Gemini & Perplexity. Data weighted by Google Search volume.]

The takeaway: citations are relatively rare, but when they happen, they tend to happen on the
queries with the most eyeballs.

## Why unlinked mentions still matter

That doesn’t mean unlinked mentions are worthless. Far from it.

Think about how LLMs actually work: they learn by reading the web. Every time your brand name
appears on a credible site connected to a specific topic, that becomes another training example.

*(screenshot)*

There’s real data behind this. In an [Ahrefs study of 75,000 brands](https://ahrefs.com/blog/ai-overview-brand-correlation/), branded web mentions had
the strongest correlation with AI visibility: a 0.664 correlation with showing up in AI overviews.
That’s stronger than backlinks, domain rating, referring domains, or any other traditional SEO
metric.

[Factors that correlate with brand appearance in AI overviews. Branded web mentions (0.664), Branded
anchors (0.527), Branded search volume (0.392), DR (0.326), # of ref domains (0.295), Branded
traffic (0.274), Number of backlinks (0.218), Ad traffic (0.216), Ad cost (0.215), URL rating
(0.18), Number of site pages (0.17)]

So even when AI doesn’t give you a link, just being mentioned builds your brand’s presence in the
model. It’s the long game, and it compounds over time.

## Not all AI responses look the same

There’s another layer to this: the type of response AI gives matters for your business.
* Step-by-step guides, for queries like "how to fix a leaky faucet" or "how to set up Google
  Analytics." If you’re a service-based business or create how-to content, these are opportunities
  if you can get AI to recommend you as an expert.
* Direct answers, for queries like "what’s the capital of France?". If you’re a publisher or
  informational brand, being the source AI cites for factual answers can build massive authority,
  though you probably won’t get a click since AI gives the answer directly.
* Video citations, covered in the previous lesson. YouTube is one of the most cited domains across
  AI platforms. If you’re a creator, your videos can show up directly in AI responses, and some
  people prefer watching video over reading text.

## AI visibility is a spectrum, not a switch

You’re not just visible or invisible. AI visibility is a spectrum, and the type that matters most
depends entirely on your business.
* If you sell a product, you want your brand showing up when people ask about your category. A query
  like "best golf balls" surfaces tons of different brands and models, and someone can click through
  from the AI response, land on a shopping result or the manufacturer’s site, and buy something
  right away.
* If you’re a publisher or content creator, training data visibility is your long game. Every
  article AI cites, every video it surfaces, reinforces your authority in the model.
* If you’re not yet visible at all, that’s the gap this course is meant to help you close.

## Checking your own AI visibility

To see where you stand in AI search, go to [Ahrefs Brand Radar](https://ahrefs.com/brand-radar) and search for your website.
You’ll immediately see how you stack up against competitors across different AI platforms.

*(screenshot)*

You can filter by:
* Mentions: when your brand is named but not linked
* Citations: when AI actually links to your site
* Impressions: how many people could potentially see your brand in AI
* AI Share of Voice: how often your brand is mentioned compared to your competitors

The gap between your impressions and your mentions is your opportunity. If AI is responding to
queries about your topic but not naming you, that’s exactly where you need to focus, which is what
module two covers: AEO strategy, research, targeting, and gaps.


### Further reading

Check out some of the articles on the [Ahrefs blog](https://ahrefs.com/blog/category/ai-search/) to learn about AEO:
* [Guide to Answer Engine Optimization](https://ahrefs.com/aeo)
* [Answer Engine Optimization: How to Win in AI-Powered Search](https://ahrefs.com/blog/answer-engine-optimization/)
* [Generative Engine Optimization: Growth Strategies and Metrics For the AI Era](https://ahrefs.com/blog/geo-generative-engine-optimization/)
* [Do AI Assistants Link When Mentioning Brands? For Ahrefs, Only 28% of the Time](https://ahrefs.com/blog/ai-citations-vs-impressions-study/)
* [An Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)](https://ahrefs.com/blog/ai-overview-brand-correlation/)

*(Source: https://ahrefs.com/academy/aeo-course/lesson-1-3)*



---

# Module 2. AEO Strategy


## Lesson 2-1: How to Run a Brand Gap Analysis

This lesson covers how to run a brand gap analysis.

A brand gap analysis measures the difference between where your brand should be showing up and where
it actually is, in Google search, AI results, and across the web generally.

It examines everything shaping your discoverability and reputation, from how AI describes you to
which competitors are cited instead of you.

## Step 1: Map out your branded entities

Your brand gets referred to differently by different people, so you want to map out:
* Your main brand name
* Any sub-brands
* Product names
* Proprietary features
* Proprietary metrics
* Personal brands associated with your company

For example, at Ahrefs, this would mean mapping out the main brand Ahrefs, product brands like [Site
Explorer](https://ahrefs.com/site-explorer), [Brand Radar](https://ahrefs.com/brand-radar), and [AI Content Helper](https://ahrefs.com/ai-content-helper), proprietary metrics like Domain
Rating and Traffic Potential, and personal brands like Tim Soulo, Patrick Stox, Ryan Law, and Glenn
Allsopp.

*(screenshot)*

Each of these has its own visibility profile in search and AI results, and the analysis in this
lesson can be repeated for each one.

### Connect Entities to Topics and Attributes

Once you have your list, connect each entity to the topics and attributes people should associate
with it.

Search engines and LLMs don’t understand brand names on their own. They infer meaning from how your
brand is described and discussed, so you need to clarify what problems you solve, what qualities
you’re known for, and what context you belong in.

A quick way to do this is keyword research: look for recurring adjectives, modifiers, and
descriptive phrases people use alongside your brand or category, things like "affordable,"
"AI-powered," or "enterprise grade," whatever applies to your space. This gives you your benchmark
for what your brand should be known and found for.

*(screenshot)*

## Step 2: Run your audit in Ahrefs

Start by entering your brand’s website into Site Explorer. You’ll get a dashboard of baseline
metrics: domain rating, referring domains, organic keywords, organic traffic, and traffic volume.
This gives you your traditional SEO snapshot.

*(screenshot)*

What matters most for AEO, though, are the AI citation metrics, which give you a snapshot of your
brand’s visibility in AI search across different platforms.

Clicking into any of those takes you to the Brand Radar report for that platform, which is where the
real audit happens.

*(screenshot)*

There are four key things to pay attention to in Brand Radar:
* Mentions: the number of times your brand is mentioned in AI responses
* Citations: the number of times your website is actually cited as a source
* Impressions: an estimation of exposure based on how often responses containing your brand are
  shown
* AI share of voice: how often your brand is mentioned compared to your competitors

### Using Brand Radar’s filters

The real power of Brand Radar is in its filters, which let you isolate the exact scenarios that
matter:
* Filter for prompts that mention your brand name in the AI platform of choice
* Filter for responses that mention your brand but don’t cite your website (missed citation
  opportunities)
* Add a competitor and filter for queries where they’re mentioned but you’re not
* Look at specific topics to see whether AI associates them with your brand or someone else’s

*(screenshot)*

## Step 3: Identify your gaps

When looking at this data, it helps to think about your gaps across six dimensions. This is a
framework that Despina from Ahrefs laid out in her [brand gap analysis guide](https://ahrefs.com/blog/brand-gap-analysis/).

[Ahrefs' six core dimensions of a brand gap analysis include visibility, narrative, format, topic,
web mention and demand gaps.]

1. Visibility gap: your brand appears less often than competitors in search or AI results.

2. Narrative gap: how AI or media describes your brand versus how you actually want to be
positioned. For example, maybe you’re a premium tool, but AI keeps calling you a budget alternative,
or your key differentiator just isn’t coming through.

3. Topic gap: topics you should be associated with but aren’t. For example, if you’re a project
management tool but AI never mentions you when people ask about remote team collaboration, that’s a
topic gap.

4. Format gap: AI tends to cite certain types of content such as guides, videos, reviews, and
comparison pages. If you’re not producing them, that’s a gap. As covered in module one, YouTube is
one of the most cited domains, so if competitors have YouTube content getting cited and you don’t,
that’s a content format gap.

5. Web mentions gap: external sources such as listicles, review sites, forums, and publications that
mention competitors but not you. As covered in module one, these third-party mentions are one of the
strongest signals for AI visibility.

6. Demand gap: branded queries or searches that signal awareness opportunities you haven’t captured.
People are searching for things in your space, but your brand name never comes up alongside those
searches.

When you look at all six together, you get a complete picture of your brand’s visibility in AI. Nt
just how often you show up, but whether you’re showing up for the right things in the right way, in
the right places.

## Step 4: Prioritize

You’re going to find a lot of gaps, and you can’t fix everything at once, so you need to prioritize
your efforts. You’ll generally do one of three things:
* Fix: improve or optimize something that already exists
* Build: create new content or pages for opportunities you’re not covering at all
* Influence: strengthen your offsite visibility through outreach and brand mentions

Start with the quick wins. If you have a page that’s already ranking but just needs a content update
to close a topic gap, that’s low effort and potentially high impact.

If you’re missing from a major listicle that all your competitors are on, that’s a web mentions gap
you can close with outreach.

As you go through this process, you’ll probably find that most of your gaps fall into one or two of
the six buckets.

### Auditing your competitors

You can run this exact same audit for your competitors to get a wealth of insights on new topics and
gaps that will be relevant to your business. Look at their branded queries to see what users and AI
connect to them more strongly than they do to you.

*(screenshot)*

Sometimes those are quick wins: you might already have the features, you just haven’t created the
content that connects them to your brand, or you don’t have enough pages talking about that feature.

A template is available to help organize all of this and share it with stakeholders. [Download
template](youtube.com/redirect?event=video_description&redir_token=QUZZTVljRy1wR0ZfTFNGWHFfWi1YOEtvck5mOHxBTl9pYzRjZmR6TGctdVJEYV81ZmtZWlRjcVBIQVpXa3lGa0xTQ0pTMkFlWHBhb21FalhqS2xmcnB0UnkwVENMSnlwMXhTSXI1cGsxUGFtNnhhMzY1Nk9pbjl2eUcxLUFCbzBh&q=https%3A%2F%2Fdocs.google.com%2Fdocument%2Fd%2F1z16ENB8G7kshEyJfXXSRQyqNVYsl7x-YRDcYWtp5vpQ%2Fcopy&v=E0VFNbZtZKM)


### Further reading

Check out some of the articles on the [Ahrefs blog](https://ahrefs.com/blog/category/ai-search/) to learn about AEO:
* [Guide to Answer Engine Optimization](https://ahrefs.com/aeo)
* [Answer Engine Optimization: How to Win in AI-Powered Search](https://ahrefs.com/blog/answer-engine-optimization/)
* [Generative Engine Optimization: Growth Strategies and Metrics For the AI Era](https://ahrefs.com/blog/geo-generative-engine-optimization/)
* [Brand Gap Analysis: Find Out Why You’re Invisible in AI Search](https://ahrefs.com/blog/brand-gap-analysis/)

*(Source: https://ahrefs.com/academy/aeo-course/lesson-2-1)*


## Lesson 2-2: Keyword & Prompt Research for AI SEO (AEO)

This lesson will help you find the actual keywords and prompts to go after. There’s a ton of overlap
between SEO and AEO, and half the battle is knowing which is which and how to approach each one
differently.

## Step 1: Build your keyword list

This part hasn’t changed much from traditional keyword research. You need two things:
* Seed keywords: broad terms related to your niche
* Modifiers: add-ons like "best" or "how to" that turn those seeds into real searches

A quick way to generate both is to ask your AI assistant of choice something like:

"I’m doing keyword research for [my type of site], which makes money through [my revenue model]. My
target audience is [this group]. Give me 10 seed keywords that are one to two words max and five
plus modifiers that will help me surface appropriate content formats I can use in my keyword
research. The seeds and modifiers should not share the same words."

This gives you a solid starting list of seeds and modifiers.

To turn that into real keyword data:
1. Paste your seed keywords into [Keywords Explorer](https://ahrefs.com/keywords-explorer)
2. Go to the Matching terms report
3. Add your modifiers using the Include filter

*(screenshot)*

You should now have hundreds, maybe thousands, of real keyword ideas your audience is actually
typing into Google.

## Step 2: Vet those keywords (the BID formula)

Some keywords look enticing (high volume, high traffic potential, low difficulty) but aren’t
actually worth targeting. Before committing to any keyword, run it through three tests, the BID
formula.
* B: Business potential. Ask yourself if ranking number one for this keyword would actually help
  your business. "What is espresso" has solid volume and low difficulty, but the searcher isn’t
  looking to buy anything, maybe ever. "Best espresso machine under $500" shows intent and a budget.
  Always choose keywords that move the needle.
* I: Intent. Google the keyword and look at what’s actually ranking. If every top result is an
  e-commerce page and you’re trying to rank a blog post, it’s not going to happen. The SERP tells
  you what searchers want, so match the intent or move on.
* D: Difficulty. Choose keywords you actually have a chance at ranking for. Check the referring
  domains and Domain Rating of the top-ranking pages. Generally, the more links and the higher the
  DR, the tougher the competition. If you see a few low-DR sites in the top 10, that’s usually a
  good sign.

If a keyword passes all three tests, you should consider targeting it, unless it fails the next
filter.

## Step 3: Apply the AI filter

Before committing to a keyword, ask one more question: can AI fully satisfy the user for this query?
Even if a keyword passes BID, if the AI overview answers it so well that there’s no reason to click
through, that keyword might be a trap.

There’s [Ahrefs data](https://ahrefs.com/blog/ai-overview-triggers/) behind this:
* AI overviews appear on about 21% of all keywords
* For informational queries, that rises to nearly 58% for question queries
* 46% of queries with seven or more words trigger an AI overview
* 99.9% of keywords that trigger AI overviews are informational in intent

So before committing to a keyword, Google it, look at what shows up, and put yourself in the
searcher’s shoes. Ask: "Am I satisfied with this answer, or do I need to click somewhere to learn
more?" If the AI overview nails it, that keyword might not be worth targeting the traditional way.

Tip

Even though AI is eating clicks for many informational keywords, there’s a whole category of
searches it hasn’t touched: free tools.

Searches like "backlink checker," "mortgage calculator," and "word counter" show no AI overview,
because when someone searches for a tool, they need to actually use something, and AI can’t replace
that (yet).

To find these opportunities:
1. Go back to Keywords Explorer with your broad seeds.
2. Head to the Matching terms report.
3. Add modifiers like calculator, checker, generator, tool, template, finder, planner, and maker.

*(screenshot)*

## Step 4: Find your AI mention opportunities

Just because an AI overview nails the answer doesn’t mean you ignore that keyword. It just needs to
be targeted differently, which is where AEO comes in.

Instead of trying to rank and earn a click, your goal becomes getting your brand mentioned in the AI
response itself.

To do that, you need to know which queries to focus on and which pages AI is pulling from. An Ahrefs
research shows that [43.8% of all cited pages (across AI overviews and ChatGPT) are listicles](https://ahrefs.com/blog/best-lists-research/).

These pages likely show up so often because they help AI build consensus: if your brand is mentioned
across multiple lists, that’s multiple sources recommending you, and AI picks up on that.

To find the queries that matter for your brand:
1. Go to[ Brand Radar](https://ahrefs.com/brand-radar) and enter your website.
2. Find your brand, hover over the AI platform you want to research, and click "Others only." This
   shows the mention gaps where competitors are showing up but you are not.
   
   *(screenshot)*
3. Filter for queries containing best, top, versus, review, or alternative.
   
   *(screenshot)*

This surfaces the queries where AI is mentioning your competitors but not you, and the ones AI is
most likely pulling from when looking for brands in your space. This is your short list.

Tip

Keep in mind that over [45% of citations change when AI overviews refresh](https://ahrefs.com/blog/ai-overview-change/), and that happens on
average every 2 days. This isn’t a one-time exercise, it’s worth revisiting this report regularly to
catch new queries as they come in.

[Ahrefs research findings for 43,000 keywords. Title: AI Overviews change every 2.15 days. Image
shows two cartoon calendars side by side. The first reads "Nov 1st" and shows an AI Overview for
"renters insurance". The second reads "Nov 3rd" and shows a different, longer AI Overview for
"renters insurance". An arrow points from the first calendar to the second, with text reading "2.15
days"]

## Step 5: Prompt research

When someone opens ChatGPT, Google AI mode, or Perplexity, they’re not typing keywords, they’re
having conversations.

They say things like "I’m a small agency owner looking for a marketing platform, which one should I
choose?" or "What’s the best way to track rankings if I’m just getting started?" They use natural
language with full context, and every person phrases it differently.

This means you can’t approach prompt research the way you approach keyword research, where you find
exact terms and go after them one by one. With AI, the same question asked 10 different ways can get
10 different answers with 10 different brands mentioned.

On top of that, AI fans each prompt out into many sub-queries behind the scenes, most of which have
zero search volume and will never repeat.

If you’re invisible for a topic, the fix isn’t optimizing for one specific prompt, it’s building
visibility across the entire topic. That’s exactly what’s covered in module three.


### Further reading

Check out some of the articles on the [Ahrefs blog](https://ahrefs.com/blog/category/ai-search/) to learn about AEO:
* [Guide to Answer Engine Optimization](https://ahrefs.com/aeo)
* [Answer Engine Optimization: How to Win in AI-Powered Search](https://ahrefs.com/blog/answer-engine-optimization/)
* [Generative Engine Optimization: Growth Strategies and Metrics For the AI Era](https://ahrefs.com/blog/geo-generative-engine-optimization/)
* [Do Self-Promotional “Best” Lists Boost ChatGPT Visibility? Study of 26,283 Source URLs](https://ahrefs.com/blog/best-lists-research/)
* [AI Overviews Change Every 2 Days (But Never Change Their Mind)](https://ahrefs.com/blog/ai-overview-change/)
* [What Triggers AI Overviews? 86 Factors and 146 Million SERPs Analyzed](https://ahrefs.com/blog/ai-overview-triggers/)

*(Source: https://ahrefs.com/academy/aeo-course/lesson-2-2)*



---

# Module 3. AEO Execution


## Lesson 3-1: How to Optimize Content for AI Search Engines

*(Transcript, this lesson is video-only on the course page)*

Hey, it’s Ammo and welcome to the third module in our AI search course. In modules one and two, we covered how AI search works and how to find the keywords and prompts to go after. Now it’s time to actually execute and that’s what this module is all about. And in this first lesson, we’re going to be talking about creating content that gets cited and I’ve got some data on what AI actually cites to back our strategy.

First, content length doesn’t matter. We analyzed over 174,000 pages cited in AI overviews and the correlation between word count and getting cited is just 0.04. That’s basically zero. Over half of all cited pages, 53.4% are under 1,000 words.

So, if you’ve been writing 3,000 word articles because you think longer content ranks better, that doesn’t apply here with AEO. AI doesn’t care how long your page is. It cares whether your page answers the question. Second, freshness matters a lot.

Content cited by AI is on average 25.7% fresher than what ranks in traditional organic results. And when you look at ChatGPT specifically, 89.7% of its top cited pages were updated in 2025. And get this, 76% were refreshed within the last 30 days. Think about that for a second.

If your content hasn’t been touched in 6 months, you’re already at a disadvantage for AI citations. Now, this doesn’t mean you should just change the publish date and call it a day. Google can detect that. You need to make meaningful updates to the actual content.

We’ll talk about how to do that later in this lesson. And the third data point is that format matters. If you remember from the last module, we talked about how 43.8% of all cited pages by ChatGPT are listicles. Best X, top X, comparisons, reviews.

That’s not a coincidence. These formats are built for AI because they give clear structured recommendations that AI can pull from. But lists aren’t the only format that work. Data-driven content with original stats also get cited heavily because AI loves citing specific numbers.

And comparison content like X versus Y pages perform well because they map directly to how people ask AI questions. So, the key takeaway here is simple. Keep your content fresh, keep it focused, and lean into formats that AI can actually use. Now, let’s talk about how to actually structure your content so AI can use it.

And here’s something a lot of people get wrong. They think they need to write differently for AI, like there’s some special AI-optimized format they need to follow. There isn’t. You’re writing for humans.

AI is trained on what humans find valuable. So, content that serves human readers well is content AI wants to cite. That said, there are a few principles that help both humans and AI get more out of your content. The first one is BLUF, which stands for bottom-line-up-front.

This one’s borrowed from military communication, and it’s simple. Start every section with the answer, not the backstory. For example, instead of opening a section with "Over the past few years, link-building strategies have evolved significantly due to changes in how search engines evaluate link quality," you could write, "The most effective way to build backlinks in 2026 is to create original research." Now, here’s why this matters. When humans scan a page, they follow what’s called an F pattern.

They read the beginning closely, skim the middle, and maybe pick up again at the end. LLMs show a similar pattern. They weigh the beginning and end of a passage more heavily than the middle. So, if your key point is buried three paragraphs into a section, both humans and AI might miss it.

So, put the answer first, then support it with context and examples. The second principle to apply is atomic content. This means every section on your page should be able to stand on its own. So, when editing your content, ask yourself, if AI pulls just one section from your article, does it still make sense?

Or does it depend on context from three paragraphs earlier? This matters because AI systems chunk your content into pieces when they process it. Different AI models chunk content differently, and you can’t control how they do it. But, if every section is self-contained, it doesn’t matter where the chunks fall because the meaning survives.

A good test you can do is to take any H2 section on your page and read it completely out of context. If it doesn’t make sense without the rest of the page, rewrite it so it does. The third principle is entity-rich writing. AI understands text by looking at entities and the relationships between them.

Entities are things like brands, products, people, places, and specific concepts. So, instead of writing, "This tool helps with SEO." write, "Ahrefs Keywords Explorer helps you find keywords with low difficulty and high traffic potential." By providing more entities, relationships, and context, AI has more to work with. And this also ties into how AI picks what to cite. The more specific and concrete your content is, the more useful it is when AI is trying to answer a specific question.

And the fourth principle is to keep it simple and declarative. Use short sentences and clear subject-verb-object structure. Basically, try to keep things to one idea per sentence. This isn’t about dumbing down your content, it’s about making it easy to parse for both people and AI.

If a sentence takes two reads to understand, it’s probably too complex. Simplify it. Your readers will thank you, and AI will have an easier time extracting the key points. Now, these four principles will help you create content that’s more likely to get cited, but there are a couple more things worth knowing that can push your chances even further.

And the first is something most people don’t think about. LLMs have a tendency to flatten originality. Like, if you come up with an original concept or framework and you’re the only one talking about it, AI will often absorb the idea without crediting you. It becomes part of its general knowledge.

The way around this is to label your ideas with your brand name. For example, instead of calling something a content scoring matrix, call it the Ahrefs content scoring matrix or your brand content scoring matrix. Define it explicitly and then distribute it widely across your blog, social media channels, podcasts, Reddit, etc. The more places it shows up with your name attached, the harder it is for AI to flatten it into generic knowledge.

Now, another thing you can do is refresh what I call sleeper pages. These are pages on your site that used to rank well, but have slowly declined over time. They already have the backlinks. They already have the authority.

They just need a refresh. And because freshness is such a strong signal for AI citations, updating these pages can be one of the fastest ways to gain AI visibility. Here’s how to find them. Go to Site Explorer and enter your domain and head to the top pages report.

Then sort by traffic change and look for pages with significant declines. What you’re looking for are pages that have two things. A decent number of backlinks, which means they already have authority, and a clear traffic decline, which usually means the content has gone stale. Now, before you update anything, make sure it’s actually a content issue and not a backlinks issue.

If the page never had many referring domains to begin with, updating the content probably won’t help. But, if it has links and the content is just outdated. That’s a high potential opportunity. So now you know how to create and update content that AI actually wants to cite.

But creating great content is only half the equation. You also need to get mentioned on other people’s content, the pages AI is already pulling from. And that’s exactly what we’ll cover in the next lesson. I’ll see you there.

*(Source: https://ahrefs.com/academy/aeo-course/lesson-3-1)*


## Lesson 3-2: Brand Mentions for SEO: How to Get Cited by AI (3 Tiers)

*(Transcript, this lesson is video-only on the course page)*

Hey, it’s Ammo and welcome to the second lesson in this module, which is on earning mentions and citations. Now, in the last lesson, we talked about creating content that gets cited. But, here’s the thing. A lot of AEO isn’t about what’s on your site, it’s about getting mentioned on other people’s pages.

And not just any pages, the ones AI wants to pull from. Branded web mentions are so important that in our study of 75,000 brands, they had the strongest correlation with visibility in Google’s AI overviews, stronger than backlinks, referring domains, and even domain rating. So, in this lesson, I’m going to show you the three types of pages you should be getting mentioned on and how to find the ones that matter most. Let’s get started.

So, I like to think about mention sources in three tiers. And the reason I use tiers is because not all mentions are created equal. Where you get mentioned matters just as much as how often you get mentioned. According to our data, we found that getting mentioned on highly linked pages has a 0.7 correlation with appearing in Google’s AI overviews.

That’s even higher than the general branded mentions correlation. So, the quality of the page you’re on makes a huge difference. Now, let’s talk about tier one mentions, which are found in third-party editorial content. These are the hardest mentions to earn, but they’re also the most valuable.

Think industry publications, review sites like Wirecutter or TechRadar, listicles and comparison posts on authoritative blogs, and YouTube reviews from creators in your space. Why are these so powerful? Because they’re exactly the kind of pages AI loves to cite. Remember, 43.8% of ChatGPT citations are listicles and comparison pages.

So, if you’re mentioned on those pages, you’re in the pool of content AI is already pulling from. Now, to find the right pages to go after, go to BrandRadar and enter your domain. Then, open the cited domains report. This shows you the top websites that AI is citing for topics related to your brand.

You’ll see the number of AI responses, the number of pages, and how often each brand is being mentioned. From here, you can identify which sites matter most for your niche. Maybe it’s kbb.com for cars, or CNET for tech, or specific niche blogs in your industry. These are the sites where getting a mention can directly lead to AI visibility.

Now, it’s important to note that you don’t have to wait for AI to start citing a page before you try to get on it. Instead, look for pages that already have a lot of links and cover your topic. Even if they’re not cited now, there’s a good chance that they’ll be cited at some point. You can use Content Explorer for this.

Search for your title and add title:best or title:versus along with a minus operator before your brand name. This will surface listicles and comparison posts in your niche where your brand isn’t mentioned. Then filter for pages with a good number of referring domains. Next up are tier two mentions, which can be found on user-generated content and community platforms.

This includes Reddit, Quora, niche forums, and community sites. And these matter more than you might think. Reddit is one of the most frequently cited sources by ChatGPT. It’s also one of the foundational training sources for large language models.

The play here isn’t to spam Reddit threads with your brand name. That’ll backfire fast. Instead, find threads where people are asking questions your product or expertise can genuinely answer. If someone’s asking, "What’s the best tool for X?" and your product is actually a good fit, contribute a real answer.

You can find these threads through Brand Radar as well. Look at the cited domains report and see if reddit.com shows up for your space. If it does, dig into which specific threads AI is pulling from and join the conversations that are already happening. Now, if you want to find niche-specific Reddit pages that are already ranking well in Google, you can go to Site Explorer, enter reddit.com, and open the organic keywords report.

Filter for keywords ranking in the top five, and add your niche terms to the include filter. And now you’ve got a list of Reddit pages that are already ranking well in Google and are related to your space. And you can do the exact same thing for niche forums and Q&A sites. Wherever your target audience is having conversations about problems that you solve, that’s where you want to be present.

Finally, we have tier three, which are your own properties. Some brands own multiple domains. Ahrefs, for example, has detailed.com and bloggerjet. If these properties are authoritative in their own right, they can serve as additional citation sources for AI.

Now, even though you don’t have the resources to run multiple sites, the principle still applies. Your YouTube channel, your podcast, your LinkedIn content, these are all indexed and can all show up as sources AI pulls from. Basically, the more places your brand appears in a positive, topically relevant way, the more training examples AI has to learn from. Now, earning mentions is one thing, but you also need to keep track of them because mentions can disappear, pages get updated, lists get refreshed, and your brand can get removed without you ever knowing.

And on the flip side, sometimes AI picks up wrong information about your brand from outdated or inaccurate sources. So, it’s worth doing a regular mention audit. In Brand Radar, you can track your overall mention trends over time. Look for any drops in mention volume.

If you find a page that was previously mentioning you, but your brand was removed during an update, that might be worth investigating. And pay attention to sentiment, too. Are AI responses about your brand accurate? Are they positive?

If you spot misinformation, your best move is to update your own content first, and then reach out to the publisher to request a correction. The faster you fix inaccurate mentions, the less time AI has to learn from them. So, between this lesson and the last one, you’ve got a clear playbook for earning AI visibility through content and mentions. But, there’s one platform we haven’t talked about yet that deserves its own lesson, and that’s YouTube.

I’ll see you in the next lesson.

*(Source: https://ahrefs.com/academy/aeo-course/lesson-3-2)*


## Lesson 3-3: YouTube SEO for AI Search: How to Rank your YouTube Videos

*(Transcript, this lesson is video-only on the course page)*

Hey, it’s Ammo and welcome to lesson three, which is on YouTube optimization for AI visibility. Now, in the last two lessons, we covered how to create content that gets cited and how to earn mentions on other people’s pages. But, there’s one platform that deserves its own conversation, and that’s YouTube. Why?

Because YouTube is the most cited domain in Google’s AI overviews. And get this, according to our data, YouTube mentions have a 0.737 correlation with ChatGPT visibility. That’s the strongest correlation of any factor we studied. And there’s a reason for that.

GPT-4 was trained on over a million hours of YouTube transcripts. So, YouTube isn’t just a platform AI sites, it’s a platform AI learns from. So, in this lesson, I’m going to walk you through a three-step process to get your YouTube videos in front of AI. Let’s get started.

So, the first step is to find what’s already working on YouTube in your niche. And the key idea here is that you want to target what I call search hits, rather than viral hits. A viral hit can get you a spike of views, and it can keep spreading based on a YouTube user’s interests. But, once the YouTube algorithm has exhausted the interested people, it dies.

A search hit, it gets you consistent traffic from both Google and YouTube search month after month because people are actively searching for that topic. And search hits are exactly the kind of videos AI is likely to pull from. And that’s because if Google is already ranking a YouTube video for a keyword, there’s a good chance AI overviews will cite it, too. On top of that, titles for search videos tend to be very clear about what the video is about, whereas viral hits, not so much.

So, here’s how to find these topics. Go to Site Explorer and enter www.youtube.com/watch. Then, open the organic keywords report. This shows you every keyword that YouTube videos are ranking for in Google.

Now, filter for keyword rankings in the top three and add your niche terms to the include filter. What you’ll get is a list of topics where YouTube videos are already ranking at the top of Google and are related to your space. Now, let’s move on to the second step, which is to actually create videos that rank. Once you found your topics, you need to make sure your video is set up to rank in Google.

And there’s a checklist I follow for this. First, your title needs to contain the keyword people are searching for. This isn’t the place for clever or click-baity titles. If the keyword is how to use Google Docs, that should be right in your title.

Save the creativity for the thumbnail. The title handles the keyword, the thumbnail sells the click. Second, your description needs to be a real summary of the video. Just write a summary of what your video covers and add your target keyword in the first couple of lines.

Google reads this, AI reads this, and viewers read it, so make it count. Third, add timestamps. Timestamps turn into YouTube chapters, and those chapters can show up in Google for specific queries. So, if your video covers five tips and someone searches for tip number three, Google can link directly to that chapter.

It’s basically free extra visibility for like two minutes of work. Fourth, say the keyword in your video. And this one’s important. Google understands audio.

Liz Reid, who’s VP of search at Google, has said that Google can understand audio content and video content. So, if your video is about the best protein powder for repair, actually say those words in the video. Don’t put it in the title and hope for the best. And fifth, match the format to what’s already ranking.

If tutorials are dominating the search results for your keyword, make a tutorial. If listicles rank, make a listicle. Don’t fight the format. Look at what’s working and match it because you’re matching the intent of the searcher.

Now, none of this is complicated, but it’s the difference between a video that gets buried and a video that shows up every time someone searches for that topic. And that brings us to the third step, which is to layer in AI visibility. So, at this point, you’ve got a video that’s optimized for Google search, but here’s where you take it a step further for AEO. Go to Brand Radar and enter a popular brand or YouTube channel and go to the topics report.

Then set a filter where the domain mentioned is youtube.com. Now, you can see exactly which queries AI is pulling YouTube videos into, so go and create content around those topics. And as for how to actually rank there, it comes back to what we already covered. Pick the right keyword, make a thorough, well-optimized video, and be comprehensive without wasting people’s time.

And here’s one more thing to keep in mind. Remember, every video you publish on YouTube is potentially training data for AI models. So, even if a video doesn’t get cited right away, the content is being absorbed. The more helpful, specific, and well-structured your videos are, the more likely AI is to learn from them and eventually recommend them.

So, that covers content, mentions, and YouTube, but there’s a technical side to AEO that most people skip entirely. Things like structured data, robots.txt, and making sure AI can actually access your site in the first place. And that’s what we’ll cover in the next lesson. I’ll see you there.

*(Source: https://ahrefs.com/academy/aeo-course/lesson-3-3)*


## Lesson 3-4: Technical SEO for AI: Robots.txt, GPTBot & llms.txt Explained

*(Transcript, this lesson is video-only on the course page)*

Hey, it’s Sam Oh, and welcome to the fourth lesson in this module, which is on the technical side of AEO. Now, I know technical can sound intimidating, but this lesson isn’t about rewriting your site’s code. It’s about making sure that AI can actually access and understand your content. And while access and understand might sound rudimentary for some of you, the reality is a lot of sites are accidentally blocking AI without even knowing it.

According to our data, around 5.9% of 140 million websites are blocking GPTBot, which is Open AI’s crawler. That’s millions of sites that are invisible to ChatGPT. So, in this lesson, I’ve got six technical checks and tips for you to make sure AI can find you so that it can promote you. Let’s get started.

So, the first thing you need to check is your robots.txt file. Robots.txt is a file on your site that tells crawlers what they can and can’t access. And the thing is, it’s not just Google’s crawler you need to think about anymore. There are now dozens of AI-specific bots that crawl the web.

The main ones you should know about are GPTBot and OAI search bot from Open AI, ClaudeBot from Anthropic, and Google extended from Google. If any of these are blocked in your robots.txt, you’re asking those AI platforms not to crawl your content. And assuming they obey your rules, they sure won’t be recommending your pages then if they don’t know what’s on them. Now, you might not have blocked these bots intentionally, but a lot of sites inherit robots.txt rules from templates or old configurations.

And some platforms add blocks by default. For example, Cloudflare has a feature called instruct AI bot traffic with robots.txt. That’s now enabled by default. When this is on, Cloudflare automatically updates your robots.txt to signal that your content shouldn’t be used for AI training.

So, if your site is on Cloudflare, you could be blocking AI crawlers without even realizing it. So, the first step is simple. Go to yourdomain.com/robots.txt and look for any lines that mention GPTBot, ClaudeBot, GoogleExtended, or OAISearchBot. If you see a disallow rule next to any of those, you’re blocking that AI crawler.

You can also use Ahrefs Site Audit to check this. Run a crawl on your site and it’ll flag any robots.txt rules that might be blocking AI crawlers. Now, while we’re on the topic of files AI reads, I want to make a quick note on something you might have heard of called llms.txt. This is a proposed standard, kind of like robots.txt, but specifically designed to tell AI systems about your site.

The idea is that you create a file at yourdomain.com/llms.txt that gives AI a summary of who you are, what your site covers, and where to find your most important content. It’s useful in theory, but as of right now, no major LLM provider officially supports it. OpenAI doesn’t use it. Anthropic publishes one on their own site, but hasn’t confirmed their crawlers actually read it.

And Google hasn’t adopted it either. So, should you create one? Well, I don’t think it’ll hurt you, but I wouldn’t prioritize it over the other things we’ve talked about in this lesson. Robots.txt is still the file that actually matters most right now.

All right, the second thing to check is how your site handles JavaScript. Some AI platforms can render JavaScript and some can’t. Without getting too technical, Gemini and Copilot can render JS, while ChatGPT’s crawler does not. So, if your content relies on JavaScript to load, which is common with single-page apps and some React or Angular frameworks, ChatGPT literally can’t see your content.

It visits the page and gets an empty shell. The fix here is server-side rendering, With which means your server sends the fully rendered HTML to the crawler instead of relying on JavaScript to build the page in the browser. If you’re already doing this for SEO, you’re covered. If not, it’s worth looking into, especially if AI visibility matters to you.

A quick way to test this is to disable JavaScript in your browser and visit your own site. If the content disappears, you have a JavaScript rendering issue that’s affecting AI crawlers, too. The third thing to consider is page speed. Now, you might be thinking page speed is an SEO thing, not an AEO thing, but it can actually matter more for AI retrieval than for traditional search.

When AI systems retrieve information in real time, they’re fetching, parsing, and chunking your pages on the fly. And if your page takes too long to load, it can get dropped before it’s even scored, so it won’t be making it into an AI response, even if the content is great. The good news is that if you’ve already optimized your core web vitals for SEO, you’re most of the way there. Fast-loading pages with clean HTML benefit both Google and AI systems.

And that brings us to the fourth tip, create clean HTML structure. This one’s straightforward. AI systems parse your content by following your HTML structure. So, if your headings are logical, your sections are well organized, and your paragraphs are focused on one idea each, AI has an easier time extracting the right information.

This ties directly back to the content principles we covered in lesson 3.1, bluff, atomic content, and entity-rich writing. Those principles aren’t just about writing style, they’re about making your content technically parsable for AI. So, when you’re structuring your pages, use proper heading hierarchy, H1 for for title, H2s for the main sections, and H3s for subsections. And make sure each section can stand on its own because AI might chunk your content at any heading boundary.

The fifth tip is about schema markup. Schema markup, which is also called structured data, is code you add to your pages to help search engines understand your content. Things like article schema, FAQ page, how-to, and local business. Now, does it help with AEO?

Honestly, the evidence is mixed. There’s no confirmed data that adding schema directly improves your chances of being cited by AI. But it doesn’t hurt. And if you’re already using it for SEO, there’s no reason to remove it.

I wouldn’t spend a ton of time on schema specifically for AEO, but if you’re setting up a new page, adding the right schema types is a good habit that makes your content easier for any system to understand. All right, the sixth tip that I have for you is to optimize for AI hallucinated URLs. AI assistants sometimes make up URLs that don’t exist on your site. They’ll recommend a page to a user.

The user clicks it, and they hit a 404 error. And this happens a lot more often than you’d expect. According to our data, AI assistants send visitors to 404 pages 2.87 times more often than Google Search does. And ChatGPT is the biggest offender with about 1% of its clicked URLs leading to 404 pages.

Now, rather than letting that 404 be the end of a visitor’s browsing journey, you should either fix or optimize those pages to get more out of them. You can do that by checking your analytics for pages that are getting traffic from AI referrers, but returning a 404 status. If you spot a hallucinated URL that’s getting consistent traffic, set up a redirect to the most relevant real page on your site. That way, you’re capturing traffic that would otherwise be lost.

Now, while creating content and getting cited is a big part of AEO, it’s only part of the picture. You also need to know if it’s actually working. And that’s exactly what we’ll be covering in module four, which is all about measuring and tracking your AI visibility. I’ll see you there.

*(Source: https://ahrefs.com/academy/aeo-course/lesson-3-4)*



---

# Module 4. Measuring AEO


## Lesson 4-1: How to Track AI Traffic in GA4 and Ahrefs Web Analytics (ChatGPT & More)

*(Transcript, this lesson is video-only on the course page)*

Hey, it’s Samo and welcome to module four in Ahrefs AI search course. Now, in module three, we covered how to create content, earn mentions, and optimize YouTube videos to get mentioned and cited in AI search. And we also talked about how you can optimize for the technical side of AEO. So, now you’ve got a playbook for execution.

But, here’s the thing. None of that matters if you can’t measure whether it’s working. And that’s what this module is all about. Now, I’ll be upfront with you.

Measuring AI visibility is a lot harder than measuring results from traditional SEO. With SEO, you’ve got Google Search Console giving you impressions, clicks, and rankings. With AI search, a lot of that data either doesn’t exist or is hidden from you. But, that doesn’t mean you’re flying blind because there are three ways to track your AI visibility.

And in this lesson, I’m going to walk you through how to set each one up. Let’s get started. So, the first thing you should track is AI referral traffic to your site. This is when someone clicks a link from ChatGPT, Perplexity, Claude, or another AI platform and lands on your site.

In analytics tools like Google or Ahrefs Web Analytics, it’ll show up as a referral visit. This is helpful in knowing how many people are finding you because of AI. But, here’s where it gets tricky. Not all AI platforms pass referral data properly.

Some of them strip it out entirely, which means the visit shows up as direct traffic in GA4. And if it does that, then there’s no real way to know that the traffic came from AI. For example, ChatGPT’s source links in search results pass referral data properly. But, in-content links on paid accounts use a no-referrer attribute.

So, those visits are actually invisible. Claude tracks properly. Perplexity tracks on the web, but not on its desktop app. Copilot tracks on the web, but not on Windows.

And Grok doesn’t pass referral data at all. So, the key takeaway here is this. Use AI referral traffic to understand the general trend of which AI platforms are sending you customers and visitors. But understand that what you see in your analytics is likely an undercount.

Now, with that caveat out of the way, let me show you how to set this up. If you’re using GA4, you’ll want to create a custom channel group that isolates AI traffic. Go to admin, then data display, then channel groups. Copy your default channel group and add a new channel called AI traffic.

Then set the source to match a regex that includes chat.openai.com, perplexity, gemini.google.com, copilot.microsoft.com, claude.ai, and deepseek.com. Once that’s set up, go to reports, then acquisition, then traffic acquisition, and select your new channel group. Now you can see the traffic AI is sending you, the pages it’s directing people to, and how those visitors behave compared to other channels. Now, if you want something simpler than GA4, Ahrefs has a free tool called web analytics that does this automatically.

It has a built-in AI search channel, so you don’t have to set up any custom groups. It also separates unknown traffic from direct traffic, which GA4 doesn’t do. That means you can get a clearer picture of where your traffic is actually coming from. All right, so once you’ve got AI traffic set up, there are two things I’d pay attention to.

First, look at which pages are getting AI traffic. These are the pages AI is already recommending to users. Make sure you keep these up to date, keep them accurate, and have clear calls to action. If AI is sending people to a page that hasn’t been updated in a year, it’ll probably stop sooner than you’d like.

And second, look at your important pages that aren’t getting AI traffic. If you’ve got a key product page or a high-value blog post that’s getting zero AI referrals that should be getting it, that’s worth investigating. It could be a content issue, a crawling issue, or it could just mean AI isn’t surfacing that topic yet. The second thing to track is AI bot activity on your site.

This one’s a bit different. Instead of tracking the humans who click through from AI, you’re tracking the AI bots themselves, the crawlers that visit your site to read and index your content. And here’s something most people don’t realize. AI bots visit your pages far more often than humans do.

So, the pages they’re hitting the most are likely your strongest citation candidates. Now, there are two types of AI bots to know about. The first type is training bots like GPT bot and Google extended. These take your content and use it to train AI models.

The second type is search and citation bots like ChatGPT user and OAI search bot. These fetch your pages in real time when a user asks a question. These are the ones that can actually drive referral traffic to your site. To track bot activity, you can use server logs if you have access to them, but the easier way is through Ahrefs bot analytics, which has a Cloudflare integration that shows you exactly which AI bots are visiting your site, how often, and which pages they’re focusing on.

And this works with a free Cloudflare plan, too. Now, what you’re looking for here are patterns. If a citation bot is hitting a specific page repeatedly, that page is likely being used as a source in AI responses. And if there are important pages that bots aren’t visiting at all, that could mean they’re hard to discover, which ties back to the internal linking and site structure we talked about in module three.

All right, the third thing to track is self-reported attribution. This one’s the simplest to explain, but it might be the most important for proving the value of AEO to your team and to your clients. You see, a lot of the impact of AI visibility doesn’t show up in your analytics at all. Someone asks ChatGPT for a recommendation, they get your brand name, then they go to the browser bar, and they type it directly in there.

That shows up as direct traffic. Or, they Google your brand name after hearing about you from AI. That’s going to show up as organic search traffic. So, the only way to capture this is to ask people directly.

Add a "How did you hear about us?" question to your sign-up flow, your checkout process, or your post-purchase survey. Include options like AI assistant, ChatGPT, Perplexity, etc. And AI search, like Google AI overviews. At Ahrefs, around 3% of our conversions came from AI over the last year, based on self-reported data.

And our AI visitors convert at a much higher rate than organic search visitors. But, we would never have known that without asking. So, if you only do one thing from this lesson, add that question to an entry survey. It’s the most direct way to connect AI visibility to actual business results.

Now, these three pillars work best when you use them together. Referral traffic tells you what AI is sending to your site. Bot analytics tell you what content AI is paying attention to. And self-attribution tells you what’s actually driving revenue.

No single source gives you the full picture. But, together, they give you a pretty clear view of how AI is interacting with your brand and where you should focus your efforts. And on top of these three pillars, you’ve still got brand radar tracking your AI visibility across platforms, which is what I showed you how to do back in module two. So, now that you’ve got your analytics set up, the big question is, is all of this actually worth it?

What’s the ROI of AEO? And what should you actually be doing on a weekly and monthly basis to keep growing your AI visibility? That’s what we’ll cover in the next and final lesson. I’ll see you there.

*(Source: https://ahrefs.com/academy/aeo-course/lesson-4-1)*


## Lesson 4-2: Your AEO Strategy + Action Plan Checklist

*(Transcript, this lesson is video-only on the course page)*

Hey, it’s Sam Oh, and welcome to the final lesson in Ahrefs' AI search course. Now, throughout this course, we’ve covered everything from how AI search works to how to find the right keywords and prompts to creating and optimizing content, how you can earn mentions, and even how to set up your analytics for AEO. But, there are three questions I haven’t answered yet that are super critical to know. Is AEO actually worth it?

How do you know if you’re making progress? And what should you actually be doing on a regular basis to keep growing? So, that’s what this lesson is about. Let’s get into it.

So, let’s start with a big question. Is AEO actually worth your time? Now, I’ll be honest. If you look at raw traffic numbers, AI search is still relatively small.

According to our data, AI referral traffic accounts for about 0.25% of a site’s total traffic on average. And Google still sends about 210 times more traffic than the top AI platforms combined. So, if you’re looking at this purely from a traffic standpoint, you might think it’s not worth it. But, here’s where the story gets interesting.

At Ahrefs, our AI visitors convert at 23 times the rate of organic search visitors. And we’re not alone. Vercel is seeing 10% conversion rates from AI traffic. Tally says AI is their largest acquisition channel and helped boost their ARR by a million dollars.

Think about that for a second. The traffic volume is small, but the quality of traffic is significantly higher. And the reason for that is simple. When AI recommends you, it’s already explained to the user why you’re a good fit.

So, the traffic arrives pre-qualified. They’re not browsing, they’re ready to act. And on top of conversion quality, AI traffic is growing fast. It’s grown about 9.7 times since last year.

ChatGPT alone has grown 85% since January and now sends more traffic than Reddit or LinkedIn. So, while the numbers are still relatively small today, the trajectory is pretty clear. AI traffic is only going to grow from here. But, here’s what I think a lot of people miss about AEO.

The real value isn’t just the clickable traffic. It’s the brand awareness that happens inside the AI conversation. Every time AI recommends your product or mentions your brand, that’s an impression you never had before. And most of those impressions never result in a click to your site.

They result in someone Googling your brand name later or engaging with your brand on social media when they see it in their feed because they recognize it. So, the way I think about it is this. AEO isn’t an alternative to SEO. It’s a new layer on top of it.

And the brands that build that layer now, while it’s still early, are going to have a massive advantage as AI search continues to grow. Now, the main downside, which we talked about in the last lesson, is that it’s not perfectly measurable. But, just because you can’t measure something precisely, it doesn’t mean it’s not working. It’s kind of like brand marketing.

You can’t track every billboard to a sale, but you know it shapes how people think about you. AEO is the same way. So, how do you know if you’re making progress? Well, if you followed along in module two, you set up your brand radar baseline and did your first brand gap analysis.

Now, it’s time to go back and check how things have changed. Pull up your brand radar dashboard and look at four key things. First, AI share of voice. How have you moved relative to your competitors?

If your share has grown, your efforts are working. If it stayed flat while a competitor has grown, you need to dig into why. At Ahrefs, we’re tracking queries and prompts at scale. So, brand radar, it gives you a really good look at the overall trend.

Second, look at cited domains. Are new domains citing you that weren’t before? This tells you if your mention-earning efforts from module three are paying off. Third, topic coverage.

Have you closed the gaps you identified? Are there new topics where you’re showing up where you weren’t before? And fourth, mention sentiment. Is AI saying accurate and positive things about your brand?

This is especially important because AI doesn’t just repeat what you put on your site. It synthesizes information from everywhere. I’d recommend doing this as a quick monthly check into a deeper competitive audit quarterly. Now, while we’re on the topic of what AI says about you, there’s something important that you need to be aware of.

AI is vulnerable to misinformation, and I don’t mean it in a theoretical way. We actually tested it. We created a completely fake luxury brand, planted three contradicting sources about it across a blog, Reddit, and Medium, and then asked eight different AI platforms about the brand. The results were pretty alarming.

After the fake sources were planted, Gemini and Perplexity repeated the misinformation in 37% of their answers. They cited fake founders, fake cities, fake pricing stories, all presented as verified facts. Now, the good news is that ChatGPT was much more robust. It stayed under 7% and cited the brand’s official FAQ in 84% of its answers.

But, the point is that not all AI platforms are equally resistant to bad information. So, what does this mean for you? It means you need to fill every information gap about your brand with specific and official content. Create an FAQ that directly addresses common questions.

Use specific numbers and dates and facts, not just vague claims, because when AI has to choose between vague truth and specific fiction, it tends to choose the specific fiction. Also, monitor what AI is saying about you regularly. If you spot inaccurate information, the fastest fix is to publish content on your own site that directly contradicts it. And then work on getting the third-party source corrected.

All right, so let’s bring this all together with your AEO action plan. If you’re wondering what to do first, here’s what I’d focus on this week. Check your robots.txt for AI bot access. This takes about 5 minutes max, and it’s the most common technical blocker.

Make sure AI can crawl your site. Next, set up your AI analytics. Create the AI traffic channel in GA4 or set up HS Web Analytics. Add a how did you hear about us question to your sign up or checkout flow.

Start measuring from day one so you can see how you progress over time. Next, update your most important pages for freshness. Remember, AI content is 25.7% fresher than what ranks in traditional search. Pick your top five to 10 pages and make meaningful updates.

New stats, updated examples, current information. Next up, run your brand gap analysis. Set up Brand Radar if you haven’t already, and go through the process we covered in module two. Know where you stand before you start optimizing.

And finally, identify your top 10 mention earning targets. Use the cited domains and pages reports in Content Explorer to find the pages where getting a mention would have the biggest impact on your AI visibility. And then on an ongoing basis, do a monthly Brand Radar check to track your progress, and a quarterly competitive audit to catch bigger shifts. And just rinse and repeat.

Everything I just mentioned is covered step-by-step in this course. So, if you need a refresher on any of it, just go back to the relevant lesson and then execute. AI SEO or AEO is still early. The tools are evolving, the data’s getting better, and the opportunity is only going to grow.

So, the biggest advantage you can have right now is simply starting before everyone else does. Thanks for watching our AEO course, and I’ll see you in the next video.

*(Source: https://ahrefs.com/academy/aeo-course/lesson-4-2)*


# PART 3, SOURCE INDEX

Corpus: **1,218 video transcripts inventoried** (1,022 substantive), ~290 hours, harvested 2025, subtitles preferred, CUDA faster-whisper large-v3-turbo for the rest. Full per-video index with titles + URLs: `D:\seo-courses\index.txt`. Clean transcripts: `D:\seo-courses\transcripts\`.

| Source group | Contents |
|---|---|
| google-search-central | Official Google: Search Console training, JavaScript SEO, ecommerce essentials, Page Experience/CWV, How Search Works, SEO Made Easy, Mythbusting, Snippets, Search for Beginners |
| ahrefs-* | SEO for Beginners (full course), Technical SEO, Keyword Research, Advanced Link Building, AI SEO/AEO, Local SEO, YouTube SEO, Blogging for Business, Content Marketing, Affiliate Marketing, WordPress SEO, misc tutorials |
| semrush-* | SEO Quickstart, Essentials, Backlinks, Competitor Research, Content Marketing, LLM Visibility, Local SEO, New Era SEO, Practical SEO Advanced, SERP Mastery, 2025 strategies |
| moz-* | Whiteboard Friday, Daily SEO Fix, technical tips, One-Hour Guide |
| neilpatel-* | SEO Unlocked (7-week program), SEO 2023, keyword research, link building, local |
| diggity-* | Anchor-text webinar, link-velocity webinar, backlinks/offsite, advanced SEO (Koray topical authority), case studies, audits |
| incomeschool-* | Answer targets/snippet engineering, search analysis, Battleship method, post recipes |
| briandean-* | Skyscraper, guestographics, Moving Man, white-hat links, YouTube SEO |
| yoast-* | Basic SEO, copywriting, structured data, video SEO, explainers |
| hubspot-aeo | AEO course (5 free tools, GEO 7-step, anatomy of an AEO page, Field Notes) |
| mega-* | 47 long-form courses: AI SEO (incl. Ross Hudgens page-type research), programmatic SEO (+doorway penalty anatomy), ecommerce ($40M playbook), technical audit (live), link building (Gotch 10-yr), keyword research, local SEO (7 courses incl. 5-hr GBP), full beginner courses, ranking-2026 tactics |

**Original example file merged:** `C:\Users\stemm\Desktop\Ahrefs-AEO-Course.md` → Appendix A above.

**Known gaps:** ~88 videos had no English captions and no successful audio pull at compile time (`D:\seo-courses\needs-audio.txt`); a handful of transcripts were metadata-only stubs. Sources dated pre-2023 are flagged where advice diverges from current guidance, Part 1 reflects the current consensus.

**Disclaimer:** SEO guidance changes; verify time-sensitive claims (AI search surfaces, GBP features, update behavior) against current Google documentation. Nothing here guarantees rankings.