AEO and GEO guides: how AI search works (2026)
Guides to how ChatGPT, Google AI and Perplexity find and cite pages, what moves AI citations, and how to measure it, each built from sourced research.
Research guides
- AEO basicsWhat is answer engine optimization (AEO)? A plain guide for 2026AEO is the work of getting a brand named, recommended and cited in AI answers from ChatGPT, Google AI Overviews, Perplexity and others.
- StrategyGEO vs AEO vs SEO: what's actually different in 2026GEO and AEO are two names for one job: getting cited in AI answers. It rests on SEO, then adds fan-out queries, off-site mentions and noisy results.
- ChatGPTHow to rank in ChatGPT: a step-by-step plan for 2026ChatGPT has no fixed rankings. You raise how often it names you: let its crawler in, answer early on key pages and earn mentions on other sites.
- ChatGPTHow ChatGPT chooses which sites to cite (2026)ChatGPT cites a few of the pages its search step retrieves, mostly from OpenAI's own index on free accounts, judged by each page's first ~202 characters.
- GoogleHow Google AI Overviews and AI Mode pick sources (2026)Both cite indexed, snippet-eligible pages matching Google's fan-out sub-queries, often outside the top 10, and rarely share cited URLs.
- PerplexityHow Perplexity ranks and cites pages (2026)Perplexity searches its own 200B+ URL index, splits pages into self-contained spans, drops stale ones and reranks the rest with a cross-encoder.
- ChatGPTWhy your brand isn't in ChatGPT answers: known vs foundAI engines either know a brand from training data or find it by searching. Each needs different work, so test both before you fix anything.
- CitationsWhat moves AI citations in 2026 (and what doesn't)Getting retrieved for fan-out searches, topical relevance and off-site mentions move AI citations. Schema, llms.txt and added stats don't.
- MeasurementHow to track your brand in ChatGPT, Gemini and Perplexity (2026)Ask fixed buyer questions on each engine many times, then report the share of answers that name, recommend or cite you, with a range.
- MeasurementMeasure AI visibility without fooling yourself (2026)Report mention rate pooled over many runs, with a confidence range and a control group. Single answers and rank positions are noise.
- CrawlersAI crawlers: the complete user-agent list (2026)Most AI companies run separate bots for training, search indexing and user fetches. Block training if you want; keep search and user bots allowed.
- llms.txtllms.txt in 2026: what it does and how to write onellms.txt is a Markdown index of key pages for AI agents. Google Search ignores it, AI search bots rarely fetch it, coding agents do.
- AgenciesAEO for agencies: a weekly workflow (2026)Track 20–50 buyer questions per client, read pooled rates with ranges over four weeks, fix access gates first, and judge every fix against a control group.
Articles
- MeasurementSEO and AI visibility: what agencies should measureSearch results and AI answers need different measurements. A practical set of metrics for each, and why they belong side by side, not in one score.
- AI visibilityBrand mentions and citations are different metricsA brand can be named without being linked, and linked without being named. Treating mentions and citations as one number hides the most useful findings.
- WorkflowHow to build an AI visibility baseline for a clientA step-by-step approach to measuring where a client stands in AI answers before any work begins, so later changes can be compared fairly.