Skip to content

AI Search (GEO/AEO)

This part teaches you how to get cited by AI answer engines — ChatGPT, Perplexity, Google AI Overviews/AI Mode, Claude, Copilot — for the questions your buyers actually ask. It is the book's centerpiece because it's where the most demand is moving and where the most snake oil is sold. Everything here is ranked by evidence: what measurably moves citations first, folklore last.

GEO vs AEO vs SEO — the terms, straight

The industry uses three overlapping labels. They describe one shift:

Term Expands to What it optimizes The win condition
SEO Search engine optimization Ranking pages in Google/Bing results Your URL in the top 10
AEO Answer engine optimization Being the answer an engine extracts Your sentence quoted in the answer box
GEO Generative engine optimization Being retrieved and cited by LLM-powered assistants Your domain in ChatGPT/Perplexity citations

In practice AEO and GEO are the same discipline (this book says "GEO"), and both sit on top of SEO, not beside it: answer engines retrieve from search indexes, so if you aren't crawlable and indexed you cannot be cited, period. See How AI finds and cites for the mechanics.

What actually moves citations

The evidence-based levers, in the order we prioritize them (each gets a chapter):

  1. Crawlability by AI bots — one WAF toggle can erase you from every AI answer while your site looks fine to humans. We watched it happen: a live business at site: zero because a bot-challenge returned 429 to Googlebot, GPTBot, and PerplexityBot alike (measured, 2026-07). → AI crawlers
  2. Bing indexing — ChatGPT search runs on Bing; industry studies put the overlap between ChatGPT citations and Bing's top organic results around ~87% (external research, compiled 2026-07). Optimizing for Bing is optimizing for ChatGPT. → Bing & Beyond
  3. Answer-first, structured content — comparison tables get extracted ~81% vs ~23% for the same facts as prose; FAQ-structured pages earn roughly 3x more citations; ~44% of ChatGPT citations come from the first 30% of a page (external research, compiled 2026-07). → Content that gets cited
  4. Structured data — rich schema appears on ~61% of ChatGPT-cited pages vs ~25% of the baseline web (external research, compiled 2026-07). → Structured data for the how, Content that gets cited for the content-page types.
  5. Off-site presence where AIs retrieve — ~85–93% of AI citations point at third-party sources, not the brand's own site. Reddit alone is ~40%. → Off-site signals
  6. Measurement — a recurring query battery across the assistants, scored and trended. Everything above is a hypothesis until measured. → Auditing your AI visibility

What's snake oil

Things sold as GEO that the evidence says to skip (details and receipts in GEO fundamentals):

  • "Instant indexing" services and plugins. Google's Indexing API is officially JobPosting/BroadcastEvent-only; off-label use risks revocation. The sitemap-ping endpoint has been dead since 2023.
  • llms.txt as a chat-citation lever. ~97% of llms.txt files get zero bot requests (Ahrefs, ~137k domains) and Google says it doesn't use the file. It has one real audience — coding agents reading docs sites.
  • "AI-specific" content rewriting, tiny-chunk fragmentation, secret "AI markup", paid mentions. Google's own AI-search guidance names these as myths (documented).
  • Fake reviews, invented ratings, astroturfed forum posts. Policy violations that also fail on mechanics — engines cite forums because they're authentic. The honesty doctrine is a ranking strategy here, not just ethics.

How to read the evidence in this part

Every load-bearing claim in these chapters carries a tier:

  • Measured — we shipped it on a live property and verified the result (the case studies, with dates).
  • External research — published studies (Ahrefs, the Princeton GEO paper, industry citation analyses), dated to when we compiled them. Percentages shift; treat them as direction and magnitude, not constants.
  • Documented — vendor documentation or stated policy (Google, OpenAI, Anthropic, Perplexity).
  • Community-reported — undocumented behavior (deep-link URL schemes especially) that we or others verified by testing. The most fragile tier; always date-stamped.

The part map

Chapter What you get
GEO fundamentals The intent-vs-brand reframe, the three-surface model, the myth-killing ground truth, the prioritized lever list
Content that gets cited Answer-first page structure, the pillar + glossary + how-to cluster model, content-page schema, freshness reality
AI crawlers and crawlability The crawler roster and what each feeds, the robots.txt allow-list, verifying real access, blocking trade-offs
llms.txt — the reality check What the data says, where it's actually real, the cheap-ship pattern, the decision rule
The Ask-AI widget Verified per-provider deep-link schemes, the self-contained-prompt rule, build notes
Off-site signals The sources answer engines actually cite and how to earn presence there without astroturfing
Auditing your AI visibility The query battery, scoring rubric, gap-to-action mapping, re-audit cadence

Start with GEO fundamentals if you're new to this; jump straight to the audit if you want a baseline today.