Back to blogBlog.categories.GEO

How to Get Cited by ChatGPT and AI Engines: The 12-Point GEO Checklist (2026)

12 concrete actions to optimize your site for ChatGPT, Perplexity, Gemini and Google AI Overviews: content, structure, authority. Full GEO checklist, free audit.

Sapian MetricsJuly 25, 20266 min read

Getting cited is something you prepare for

Your customers ask their questions to ChatGPT, Perplexity or Gemini, and the answer contains only a handful of brands. Either you are one of them, or you don't exist. The good news: AI engines don't pick their sources at random. They favor content that is clear, structured, factual and corroborated by multiple sources. Every one of those criteria can be worked on.

Here is the checklist we apply, to our clients and to ourselves. Twelve points, grouped into three families: content, technical structure, authority. No need to do everything at once: measure your starting point first, then work through the list in order.

Before you start: measure your starting point

Optimizing without measuring is sailing without a compass. Before the first action, establish where you stand today: across your 20 to 30 strategic customer questions, how often is your brand cited, on which engine, and against which competitors? That is your baseline citation rate; the full method is in our guide to measuring AI visibility, and the free audit gives you that snapshot in minutes.

Family 1: content AI engines want to cite

1. Answer real questions, one per page

AI engines answer questions. Every important page on your site should map to a precise question your customers actually ask: "what is the best [service] in [city]?", "[tool A] or [tool B]?". A page that tries to say everything answers nothing.

2. Put the direct answer at the top of each section

Under each heading, give the answer in 40 to 60 words before you elaborate. AI engines extract passages, not whole pages: a self-contained, factual, dated paragraph is an ideal citation candidate. Context and nuance come after.

3. Write a canonical definition of your brand

One single sentence, reused everywhere (site, LinkedIn, directories): who you are, what you do, for whom. AI engines reconstruct their knowledge of a brand from scattered fragments; if every source describes it differently, the synthesis becomes vague or wrong. Check what the AIs already say about you: factual errors (pricing, offering, location) are common and can be fixed.

4. Give facts, figures and dates

"Market leader" is not citable. "Platform launched in 2026, 8 engines tracked, from €55 per month" is. AI engines favor verifiable claims and penalize marketing vagueness. Date your content and keep it updated: freshness is a documented selection criterion, especially on Perplexity.

5. Cover the comparisons, even the uncomfortable ones

"Alternative to [leader]", "[you] vs [competitor]": these questions are asked of AI engines every day. If you don't have a page that answers them honestly, the AIs will compose the answer from other people's sources. A factual comparison that acknowledges everyone's strengths gets cited more than a one-sided pitch.

6. Add an honest FAQ to your key pages

Five to eight questions your customers really ask, with short, complete answers. The question-and-answer format is the closest to how AI engines build their responses; it is also the easiest to extract.

Family 2: structure machines can read

7. A clean heading hierarchy

One H1 only, H2s that ask questions or clearly announce the section, short paragraphs under each. AI engines slice your pages into passages: a logical structure produces passages that stand alone and make sense out of context.

8. Mark up with structured data

Schema.org: Organization with your canonical definition, FAQPage on your FAQs, Product and Article where relevant. Markup is not magic, but it removes ambiguity about who you are and what the page says, for classic search engines and for the AIs that rely on their indexes.

9. Stay readable without JavaScript

Most AI crawlers read raw HTML and execute JavaScript poorly or not at all. If your main content only appears after client-side rendering, it is invisible to them. Test it: view the source code of your key pages; what you can't see there, the AIs can't see either.

10. Let the AI crawlers in

Check your robots.txt: GPTBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot, Google-Extended. Blocking these agents, deliberately or through an overzealous firewall, takes you out of the game. It is a strategic choice that is yours to make, but it should be a conscious one.

Family 3: authority that tips the scales

11. Be present where each engine draws from

Every AI has its favorite sources: Perplexity leans heavily on fresh pages and media, ChatGPT on Bing and on what multiple sources corroborate, Google AI Overviews on its own index. Industry directories, customer reviews, guest articles, Wikipedia pages or local listings where legitimate: every consistent mention of your brand outside your site increases the probability of citation. The engine-by-engine detail is in our analysis of citation logic.

12. Keep your SEO solid

GEO does not replace SEO, it builds on it. Google AI Overviews overwhelmingly cites pages that already rank well, and Bing feeds ChatGPT. A slow, poorly indexed site with no backlinks starts with a handicap on every engine. The two disciplines reinforce each other.

Then: re-test, or you will never know

An optimization without a re-test is an act of faith. After each wave of actions, re-test the same questions on the same engines and compare: citation rate, share of voice, factual accuracy. That is the measure, fix, prove loop; the metrics to track are detailed in our guide to core GEO KPIs. Allow 4 to 8 weeks to see the first effects: AI engines take time to re-crawl and integrate new sources.

Frequently asked questions

How do you optimize a site for GEO? By working on three families of criteria: content that directly answers real questions with verifiable facts, a machine-readable structure (clean headings, structured data, HTML accessible without JavaScript), and authority corroborated by external sources. Then by re-testing regularly to measure progress.

How long before you see results in GEO? Generally 4 to 8 weeks after the first optimizations, the time it takes for crawlers to come back and indexes to update. Factual error corrections can be faster on web-connected engines like Perplexity.

Should you optimize differently for ChatGPT and Google AI Overviews? The foundation is shared (clear content, facts, structure), but the sources differ: AI Overviews rewards your existing Google SEO, ChatGPT relies on Bing and multi-source corroboration, Perplexity on freshness. Tracking each engine separately lets you invest where the return is strongest.

What is a tracked prompt? A customer question you test regularly on AI engines to measure whether your brand appears in the answer. A good prompt set reflects your market's real questions, not just your brand name.

Does GEO replace SEO? No. GEO builds on SEO's foundations (indexing, authority, backlinks) and adds AI-specific criteria: extractability, corroboration, factual accuracy. Abandoning one for the other is the worst strategy.

Conclusion

Being cited by AI engines is neither luck nor mystery: it is the result of content designed to answer, structure designed to be machine-read, and authority corroborated beyond your own site. The checklist above is your work plan; measurement is your proof. To know where you're starting from, run the free AI visibility audit: in a few minutes you will know where you are cited, where you are absent, and which checklist item to start with.

#GEO#checklist#AI optimization#ChatGPT

Are you cited by AI?

Get a free audit of your visibility across ChatGPT, Perplexity and Google AI.

Start my free audit