Getting Cited by AI: What Matters for ChatGPT, Perplexity and Google

How to get cited by ChatGPT, Perplexity and Google's AI Overviews with the pages you already have: let the crawlers in, give the answer early, source your figures. What Google and the studies say.

Published on · Updated

By

Getting Cited by AI: What Matters for ChatGPT, Perplexity and Google
In this article

Yes, you can improve your chances of being cited by AI without publishing new content. For Google, no special optimization is needed: a page that is indexed and eligible to show with a snippet can be used in AI Overviews and AI Mode. For ChatGPT and Perplexity, their search crawlers must be able to read your pages. After that, two writing habits help your readers as much as the engines: give the answer early and back your figures with sources.

Key Takeaways

  • Google: no additional requirement for AI Overviews and AI Mode, no special file and no rewriting for AI
  • ChatGPT and Perplexity: a site that blocks OAI-SearchBot does not appear in ChatGPT search answers, and Perplexity recommends allowing PerplexityBot
  • In an analysis of 18,012 ChatGPT citations, 44.2% came from the first 30% of the text (Kevin Indig, Search Engine Journal, 2026): put the answer at the start
  • In the GEO study (Aggarwal et al., KDD 2024), citing sources, adding quotations and adding statistics were among the most effective methods; keyword stuffing did not help
  • Start with your existing pages: that is often where the most useful work is

What Google Says, and What It Does Not Ask For

Google says so in its documentation on AI features in Search: there are no additional requirements and no special optimizations needed to appear in AI Overviews or AI Mode. SEO best practices still apply. A page simply needs to be indexed and eligible to be shown in Google Search with a snippet.

Its guide to generative AI features in Search also spells out what you do not need:

  • no new machine-readable file, AI text file (such as llms.txt) or special markup;
  • no breaking your content into tiny pieces;
  • no way of writing specific to generative search;
  • no structured data required for generative AI (it remains useful for rich results).

What helps, according to the same guide: unique, useful content, a good page experience and an up-to-date Google Business Profile. To measure, Search Console offers a generative AI report.

In other words, the "first 30% rule" described below is not a Google lever. It is an observation about ChatGPT, and a good writing habit for all your readers.

ChatGPT and Perplexity: Let Their Search Crawlers In

Each engine has its own crawler, and each one is set in your site's robots.txt file:

  • OAI-SearchBot is used to show websites in ChatGPT search. According to OpenAI, sites that block it are not shown in ChatGPT search answers. GPTBot, on the other hand, is used for model training: it is a separate setting.
  • PerplexityBot is used to show and link websites in Perplexity's results, and Perplexity recommends allowing it in robots.txt.
  • Googlebot: for Google, a page must be crawled and indexed to be used in AI Overviews.

Open your site's robots.txt and look for these names. If one of them is targeted by a Disallow rule, that is the first fix to make: a blocked search crawler keeps its engine from showing your pages in its results.

Where to Put the Answer: the First 30%

Kevin Indig analyzed 18,012 ChatGPT citations to see where in the page they came from ("The Science Of How AI Pays Attention", Search Engine Journal, February 17, 2026):

  • 44.2% came from the first 30% of the text;
  • 31.1% from the middle (30% to 70% of the text);
  • 24.7% from the last third.

Two caveats. The analysis covers ChatGPT only. And the rest of the text gets cited too: the start counts more, not alone. The practical lesson is simple: whatever answers the question should come early.

How to apply it:

  1. A one-sentence definition, understandable without reading the rest.
  2. The answer before the context: the first paragraph answers the question in the title.
  3. Steps or criteria as a list, easy to reuse as they are.
  4. Nuances, examples and proof after that.

Example of a self-contained definition:

The first 30% rule means putting the definition, the answer and the key steps in the first third of a text, where Kevin Indig's analysis found the largest share of ChatGPT citations.

Sources and Quotations: What the GEO Study Measured

The study "GEO: Generative Engine Optimization" (Aggarwal et al., KDD 2024) tested text changes on an experimental generative engine built on GPT-3.5, then on Perplexity. Citing sources, adding quotations and adding statistics were among the most effective methods for visibility in generated answers. Keyword stuffing added little to nothing, and how well each method worked varied by domain.

What the study did not measure: FAQs, structured data or the length of introductions. These practices can help your readers, but the study says nothing about their effect on citations.

5 Improvements to Make on Your Existing Pages

1. Check Your robots.txt

Make sure Googlebot, OAI-SearchBot and PerplexityBot can read your important pages. The check is quick, and it decides whether the rest of the work can be seen by these engines.

2. Give the Answer in the First Paragraph

Review your important articles and service pages with one rule: the first paragraph answers the question the title asks. Not a promise of an answer: the answer itself.

3. Source Every Figure

A useful figure gives its source, its date and a link to the study or the official page. Remove the ones you cannot trace: "most businesses…" with no source tells your reader nothing.

4. Quote Real People

A quotation attributed to a named person, with their title (your lead technician, an engineer, a recognized organization), shows where the expertise comes from. Adding quotations was among the effective methods in the GEO study.

5. Answer Your Customers' Real Questions

Add a frequently asked questions section to key pages, with the questions your customers actually ask on the phone or by email. It serves your readers first; Google requires no special markup for generative AI.

Common Mistakes

  • The introduction that says nothing: "In a constantly evolving world…" teaches no one anything.
  • Explaining at length before answering: the answer arrives after the first third, or not at all.
  • Figures without a source, or buried mid-article.
  • Publishing many pages with no added value: Google says that generating many pages without adding value for users may violate its policy on scaled content abuse (Google).

How to Know Whether Your Pages Are Cited

  1. In Search Console, open the generative AI report to see how your pages perform in Google's AI features.
  2. Ask ChatGPT and Perplexity the questions your customers ask. Note the date, the sources cited and who is cited instead of you.
  3. Repeat the same checks, with the same questions, after each change.

Start with one page. Measure. Adjust.

For a business that sells in its own region, read our guide Local SEO in the AI era as well.

Conclusion

Getting cited by AI is first a matter of access and clarity: pages the crawlers can read, an answer that comes early and figures that can be traced. Google asks for no special file and no rewrite. That is the work our AI visibility service covers: crawler access, clear pages and follow-up in Search Console, ChatGPT and Perplexity.

Frequently asked questions

Do you need to publish new content to be cited by AI?

No. Start with your existing pages: make sure search crawlers can read them, give the answer in the first paragraph and source your figures. For Google, no special optimization is needed to appear in AI Overviews.

Do you need an llms.txt file or special markup for Google?

No. Google says no AI text file, no special markup and no AI-specific way of writing is needed to appear in Google Search. The llms.txt file is a proposed standard to help AI assistants use a website; it remains optional.

What is the first 30% rule?

It is the habit of putting the definition, the answer and the key steps in the first third of a text. It rests on an analysis of 18,012 ChatGPT citations, 44.2% of which came from the first 30% of the text (Kevin Indig, Search Engine Journal, 2026). It is not a Google rule.

Does blocking crawlers in robots.txt prevent citations?

For ChatGPT search, yes: according to OpenAI, sites that block OAI-SearchBot are not shown in its search answers. Perplexity recommends allowing PerplexityBot. Blocking GPTBot concerns the training of OpenAI's models, a separate setting.

Does keyword stuffing help you get cited?

No. In the GEO study by Aggarwal et al. (KDD 2024), keyword stuffing added little to nothing to visibility in generated answers. Citing sources, adding quotations and adding statistics were among the most effective methods.

How do I know whether my pages are cited?

Check the generative AI report in Search Console, then ask ChatGPT and Perplexity your customers' questions, noting the date and the sources cited. Repeat the same checks after each change.

Make your pages readable and easy for AI to cite with our AI visibility service.

Explore the service

Start with a free SEO audit

No cost, no commitment.

Get my free SEO auditBook a discovery call

We use cookies This site uses cookies to improve your experience. Some are essential for the site to function; others require your consent. Learn more.