Free playbook · updated July 2026

The DIY SEO / AEO / GEO Playbook

Everything you can do yourself to rank on Google and get cited by ChatGPT, Perplexity, and AI Overviews — 34 steps in four layers, each with its evidence attached. Tick items off as you go; progress saves in your browser.

+109%

citation lift, answer-first content

GeoSource

30–40%

visibility gain from stats & citations

Princeton, KDD 2024

~27%

of sites block an AI crawler by accident

crawler audits

3–6 mo

honest timeline for results

practitioner consensus

SEO, AEO, GEO — three layers, one pipeline

They are not alternatives; they stack. SEO gets you retrieved, AEO gets you extracted, GEO gets you cited. Full definitions: what is GEO, what is AEO, GEO vs SEO.

Layer	What it does	Where it shows	Success metric
SEO	Rank your pages in search	Google & Bing results	Rankings, clicks
AEO	Become the extracted answer	Snippets, AI Overviews, voice	Answer share of voice
GEO	Get cited in AI responses	ChatGPT, Perplexity, Gemini, Claude	Mentions & citations per prompt set

Google’s own guidance: its generative AI features run on the same ranking systems as classic Search — there is no separate AI index. Read the source.

Your progress · saved in this browser0/34 · 0%

AEO

Layer 2 — Answer Engine Optimization

0/9 done

Make every passage extractable. The unit of AEO is the passage, not the page — extractors quote sections that stand alone.

Content formatting for direct answers

Open every page with a 40–80 word direct answerDeclarative answer-first content lifted citation rates 109% in GeoSource’s 540-check study — the strongest signal measured.High impact
Phrase H2/H3 headings as real questionsHeadings should state what the section answers — they’re matched against the queries engines fan out to.High impact
Add an intent-matched FAQ to key pagesOnly questions this page’s searcher would ask next. Off-intent FAQs dilute the page’s signal.Medium
Write standalone, quotable sentencesPages with 5+ sentences that survive extraction out of context were cited ~3.2x more in practitioner testing.High impact
Use real tables for any “X vs Y” comparisonComparison and listicle formats take ~22% of all AI citations — up to ~41% on commercial queries.Medium
Keep paragraphs short — one idea eachUnder ~120 words. Long paragraphs bury the extractable sentence.Medium

Structured data (honest version)

Add Organization, Person, and Article schemaFor rich results and entity disambiguation. Honesty note: Ahrefs’ 5,885-page controlled study found no AI-citation lift from schema — do it for the right reasons.Medium
FAQPage schema only where FAQs are visibleMarkup mirrors on-page content; it never substitutes for it.Low
Show real Published and Updated datesVisible near the H1 and in schema. Never fake freshness — engines compare against crawl history.Medium

GEO

Layer 3 — Generative Engine Optimization

0/9 done

Get retrieved, then get cited. These are the measured citation drivers from the Princeton GEO research (KDD 2024) and large-scale studies.

The measured big four — every important page

Cite authoritative sources, with names and links+30–40% AI-answer visibility in the Princeton benchmark; +65% citation-quality lift in GeoSource’s checks.High impact
Include specific statistics and data pointsThe single best-tested GEO tactic (+30–40%, Princeton). “21.5% open rate” beats “high open rates” every time.High impact
Add expert quotes with attributionQuotation Addition was the third measured winner in the same research.High impact
Define key terms explicitly — “X is Y”Explicit definitions added a further +33% citation lift (GeoSource).High impact

Structure and depth

Declare one primary topic per pageOne page per intent. Ten phrasings of one question belong on one page; two questions never share one.Medium
Build interlinked topic clustersQuery fan-out turns one prompt into several searches — a complete cluster is retrievable for more of them.High impact
Publish one original data piece per quarterA survey, benchmark, or teardown. Original numbers are what earn citations and mentions — and nobody can copy them.High impact
Test your pages in ChatGPT and Perplexity monthlyFixed prompt set, repeated runs, averaged. Expect Perplexity to react in weeks; ChatGPT in 1–2 months.Medium
Refresh commercial pages quarterly — leave evergreen aloneFreshness matters most for commercial intent. The average ChatGPT-cited page is ~500 days old (Ahrefs).Medium

Entity

Layer 4 — Entity & Authority (free)

0/5 done

AI engines cross-reference sources before naming a brand. Identical facts everywhere is what makes them confident enough to mention you.

Signals you control

Write one canonical company descriptionPropagate it verbatim: site, LinkedIn, review platforms, directories. Conflicting descriptions get brands dropped from AI answers.High impact
Name every author with real credentialsSpecific expertise plus a profile link. Practitioner test: detailed author bios moved citation rates 28% → 43% in four weeks.Medium
Earn third-party mentions where engines already lookBrands are ~6.5x more likely to be cited via third-party sites than their own domain (AirOps). Reviews, communities, publications.High impact
Claim and complete review-platform profilesG2 alone is ~6–7% of ChatGPT’s B2B software citations. A thin profile is a citation that never happens.Medium
Link your schema together with sameAsOrganization/Person schema pointing at your real profiles builds a small knowledge graph engines can verify.Low

Free tools that cover all of it

Everything in this playbook is doable with free tooling. Paid trackers help at scale — our honest tool guide covers when.

Google Search Console ↗Indexing, queries, AI-experience visibility Bing Webmaster Tools ↗Bing index (feeds ChatGPT search) + AI performance reports PageSpeed Insights ↗Core Web Vitals, free Rich Results Test ↗Validate your schema Ahrefs Webmaster Tools ↗Free site audit + backlinks Knowatoa ↗Check AI crawlers can reach your site Schema Markup Validator ↗Schema.org validation Google Trends ↗Topic demand and seasonality

Your first 90 days, then monthly

Month 1 — Foundation

Search Console + Bing verification
Crawler access audit (incl. CDN “invisible 403” check)
Core Web Vitals fixes
Canonical entity description written and propagated
Answer-first restructure of your 5 highest-traffic pages

Month 2 — Content

Map your topic cluster (30–50 real questions)
Publish the pillar + first supporting pages, answer-first
Add sourced statistics and definitions to existing pages
Set up the 20-prompt AI-visibility tracking sheet

Month 3 — Authority

Publish your first original data piece
Pitch it to 20–30 niche journalists/newsletters
Claim review-platform and directory profiles
Genuine participation in 2–3 communities where buyers ask questions

Ongoing — Monthly

Run the prompt set across ChatGPT/Perplexity/Gemini, log mentions
Refresh one commercial page
One new cluster page or data update
Check Search Console AI-experience visibility + crawler logs

How to measure success

Track two things monthly: classic search (impressions and positions in Search Console) and AI visibility — the share of a fixed 20–40 prompt set where your brand is mentioned or cited. Run each prompt several times and average; single samples are noise.

The full measurement protocol, including how to read tool dashboards without fooling yourself, is in AI visibility: how to measure it. For the ranked practice list this playbook draws on, see best practices — and vote on what worked for you.