How to get cited by ChatGPT and AI search
I read what Google, OpenAI, Microsoft and the independent studies actually say, and put it in plain language.
On this page
The short answer
To get cited by AI search, do what the platforms' own guidance and the research point to: let their crawlers reach your pages, answer real questions clearly and first, back your claims with original facts and named sources, and keep your business details the same everywhere. Google says no special markup or file is needed. That is what I found after reading the sources below, and it is where I would put your effort.
Key takeaways
- Optimizing for AI search is mostly still SEO. Google's own guide says so.
- When an AI summary appears, people click less: they clicked a regular result in 8% of visits, against 15% without a summary.
- Ranking on page one helps but does not decide everything. The studies I found put the share of AI Overview citations from top-10 pages anywhere from 38% to 76%.
- Your site has to be reachable. OpenAI says sites that opt out of its search crawler will not be shown in ChatGPT search answers.
- In the one peer-reviewed test I found, adding statistics, credible quotes and citations lifted visibility by 30 to 40% on one measure.
- Google says you do not need special schema or new machine-readable files. Microsoft recommends schema. I would do the basics that help every engine.
Is AI search optimization different from SEO?
Mostly no. I started with Google's own guide, and it says so directly.
Google's guide on generative AI features, updated in July 2026, says its AI features are rooted in its core Search ranking, so the usual SEO best practices still apply. It even names the trend people call GEO and AEO, and answers it in one line. Softlixx sells SEO and AI search work, so I have a stake in this topic.
From Google Search's perspective, optimizing for generative AI search is optimizing for the search experience, and thus still SEO.
There is one real difference, and it is how an answer gets put together. Google says AI Overviews and AI Mode may run several related searches across subtopics before writing one response, a technique it calls query fan-out. A useful note on this came from Search Engine Journal's report on a newer Ahrefs study: fan-out may help explain why a page can be cited even when it does not rank for the exact question.
- Someone asks a questionOften a long, specific one.
- Related searches runThe system may search several subtopics at once (query fan-out).
- Pages are broken into piecesEach piece is weighed for authority, clarity and relevance.
- One answer is writtenWith links to the sources it used.
Simplified from Google's and Microsoft's public descriptions. Each company's real process is more complex. Sources: Google Search Central, Microsoft Advertising.
For local businesses, the same guide says a Business Profile can help products and services show up in AI responses and other Google Search features. I cover that in the research on Google Business Profile and local search.
Are AI answers really reducing clicks?
Yes, and I found two large studies that agree, even though they measure it differently.
Here is the first one. The Pew Research Center tracked the Google browsing of 900 U.S. adults during March 2025, covering 68,879 searches, 12,593 of which produced an AI summary. When a summary appeared, people clicked a regular result in 8% of visits. When it did not, they clicked in 15%. Clicking a link inside the summary itself happened in just 1% of visits, and people were more likely to end their browsing session after a results page with a summary (26%) than without one (16%).
The second one looks at the wider picture. SparkToro used Similarweb panel data for the United States from January to April 2026 and found that 68.01% of Google searches ended without a click, up from 60.45% in 2024. It also puts AI Overviews on 20% or more of all searches and says they cut click-through rates by nearly 60% when present.
That's the fastest acceleration of this phenomenon in the last decade, almost certainly driven by the massive growth in AI Overviews…
Share of Google visits by U.S. adults, March 2025 (Pew Research Center)
No AI summaryAI summary shown
Clicked a regular search result
Ended the browsing session after the results page
Clicked a link inside the AI summary itself
Bars use a scale of 0 to 30%.
Google searches that ended without a click (U.S.)
Bars use a scale of 0 to 100%. SparkToro, Similarweb panel data.
Sources: Pew Research Center, SparkToro.
My readingBoth studies measure Google in the United States, and neither can tell you what will happen to one particular website. But this is why I think being the named, linked source inside an answer is worth more than it used to be.
Do you need to rank on page one to be cited?
It helps, but the studies disagree about how much, and the answer keeps changing. I found three different numbers.
In July 2025, Ahrefs found that 76.1% of pages cited in Google's AI Overviews ranked in the top 10 for the same search (1.9 million citations from 1 million AI Overviews). Two months later, BrightEdge reported that 54.5% of AI Overview citations ranked organically, up from 32.3% in May 2024, and stressed that individual industries vary wildly.
Then I read the newest one. In March 2026, Search Engine Journal reported a newer Ahrefs analysis of 4 million AI Overview URLs across 863,000 keywords. This time 38% ranked in the top 10, 31.2% ranked between 11 and 100, and 31.0% ranked beyond 100. Ahrefs pointed to better citation detection, the fan-out process, and Google's upgrade of AI Overviews to Gemini 3 in January as context.
Share of Google AI Overview citations that also rank on page one, as each study measured it.
Ahrefs, July 2025
Cited pages ranking in the top 10. 1.9 million citations from 1 million AI Overviews.
BrightEdge, September 2025
Citations that rank organically, up from 32.3% in May 2024. Nine industries.
Ahrefs, March 2026 update
Cited pages ranking in the top 10. 4 million AI Overview URLs, 863,000 keywords.
Where the citations in the March 2026 Ahrefs update ranked
Different methods and dates, so the studies are not directly comparable. Sources: Ahrefs, BrightEdge, Search Engine Journal.
My readingThe studies use different methods and dates, so I would not compare them number for number. The plain conclusion holds either way: ranking well still raises your chances, and a page that ranks nowhere near the top can still be cited.
What kind of content gets picked up?
Clear, original, well-sourced pages. Google, Microsoft and the one peer-reviewed test I found all point the same way.
Start with what the platforms say. Google asks for non-commodity, people-first content, and says a unique point of view can help because its AI systems look at a variety of sources. Microsoft says assistants like Copilot break content into smaller, structured pieces and evaluate each for authority and relevance. It advises clear headings, and warns against hiding important answers in tabs or expandable menus, because AI systems may not render hidden content.
Now the evidence on what to write. The strongest I found is a peer-reviewed study, presented at KDD 2024 by Aggarwal and colleagues. The researchers tested several ways of rewriting web content on a benchmark of 10,000 queries. In the full paper, the three best performers were adding statistics, adding credible quotes and citing reliable sources. They lifted visibility by 30 to 40% on one measure and 15 to 30% on another. The paper lists keyword stuffing, the classic SEO trick, among its non-performing methods.
My readingThat study is a research benchmark, not a live test of ChatGPT or Google, and its authors note that results vary across domains. I treat it as a strong hint about what helps, not a guarantee.
Trust is the most important ingredient for success in organic and AI Search.
Can AI tools even see your site?
Only if you let them in. Each company runs separate crawlers, and blocking the wrong one can remove you from its answers.
I went to each company's own documentation for this one. OpenAI runs three crawlers with three jobs. OAI-SearchBot surfaces websites in ChatGPT's search features. GPTBot crawls content that may be used to train its models. ChatGPT-User visits a page when someone asks ChatGPT a question. Each setting is independent, so a site can allow OAI-SearchBot to appear in search while disallowing GPTBot. Sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, though they can still appear as plain links.
Perplexity says PerplexityBot is designed to surface and link websites in Perplexity results, and is not used to crawl content for AI foundation models. It recommends allowing PerplexityBot in robots.txt. For Google, the rule is simpler: a page must be indexed and eligible to be shown in Google Search with a snippet to be a supporting link in AI Overviews or AI Mode.
Decides if you appearModel trainingA person asked
Sources: OpenAI, Perplexity, Google Search Central.
Access problems can hide in places you would not think to look. SEO consultant Aleyda Solís has described finding that her own hosting company was blocking AI bots, which made her results look wrong until she dug into it. That story is why I would check your host and CDN settings before anything else. The speed side of that foundation is covered in what makes a website fast in 2026.
You need to allow AI crawlers to access your content.
Do you need special markup or an llms.txt file?
Not for Google. Microsoft takes a slightly different view, so I would do the basics that help both.
Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add.
Google also says you do not need new machine-readable files, AI text files, markup or Markdown to appear in Google Search, and Search Engine Journal reads that as covering llms.txt files. Microsoft, on the other hand, explains that schema is a type of code that helps search engines and AI systems understand your content, and recommends it in its guidance for AI search answers.
My readingClean structured data is cheap and harmless, and it helps some systems. I just would not expect it to be a switch that turns on citations.
How do you know if it is working?
Google and Bing now report it, and you can check the rest by hand.
In June 2026, Google added Search Generative AI performance reports to Search Console. They show impressions, pages, countries, devices and dates for AI Overviews and AI Mode, and were rolled out to all websites worldwide as of 31 August 2026. In February 2026, Bing Webmaster Tools added AI Performance, which shows how publisher content appears across Microsoft Copilot, AI-generated summaries in Bing and select partner integrations.
Changes are not instant either. For how long to wait before judging the result, see how long SEO takes, and what Google says. Whether updating a page helps it get cited is covered in the research on content freshness and AI citations.
My readingFor assistants without a report, the method I would use is simple: keep a fixed list of real questions your customers ask, ask them once a month, and note whether your business is named and linked.
What should a business do first?
If I were starting today, this is the order I would work in: access, then answers, then evidence, then consistent business details. It is my reading of everything above.
- Let the search crawlers inAllow OAI-SearchBot and PerplexityBot if you want to appear in ChatGPT and Perplexity, and check that your host or CDN is not blocking them.
- Get your key pages indexedMake sure Google and Bing have indexed them and can show a snippet.
- Put the answer firstGive the direct answer in the first two or three sentences, then the detail.
- Use headings that match real questionsClear H2 and H3 headings mark where one idea ends and the next begins.
- Add what only you knowYour own data, examples and named sources. Quote credible people accurately and link them.
- Keep your basic facts identical everywhereBusiness name, services, location and contact details, on your site and every profile.
- Measure every monthUse Search Console, Bing Webmaster Tools and your fixed list of real questions.
Questions people ask
Do I need schema markup or an llms.txt file to be cited by AI search?
Google says no. It says structured data is not required for generative AI search, that there is no special schema.org markup to add, and that you do not need new machine-readable files, AI text files, markup or Markdown to appear in Google Search. Microsoft recommends schema because it helps systems understand content. I would treat structured data as a reasonable extra, not a requirement.
Should I block AI crawlers?
It is a choice, and the settings are separate. OpenAI says a site can allow OAI-SearchBot, which is used for ChatGPT search, while disallowing GPTBot, which is used for training. Perplexity says PerplexityBot is not used to crawl content for AI foundation models. Sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, so I would decide on purpose.
Is generative engine optimization (GEO) different from SEO?
Google says optimizing for generative AI search is still SEO. The peer-reviewed GEO study found that adding statistics, credible quotes and citations improved visibility in its test, which is also good practice for ordinary SEO. So I see the two as overlapping heavily.
Will AI search reduce my website traffic?
Studies from Pew and SparkToro show fewer clicks when AI summaries appear, and more Google searches ending without a click, in the United States. No one can promise what will happen to a specific site, so I would measure your own impressions and clicks over time.
Sources
- Optimizing your website for generative AI features on Google SearchGoogle Search Central. Updated 10 July 2026.
- AI features and your websiteGoogle Search Central. Updated 10 December 2025.
- Introducing Search Generative AI performance reports in Search ConsoleGoogle Search Central Blog. 3 June 2026.
- Google users are less likely to click on links when an AI summary appears in the resultsPew Research Center (Athena Chapekis and Anna Lieb). 22 July 2025.
- In 2026, less than one third of Google searches still send a clickSparkToro (Rand Fishkin). 9 June 2026.
- 76% of AI Overview citations pull from the top 10Ahrefs (Louise Linehan). 21 July 2025.
- Google AI Overview citations from top-ranking pages drop sharplySearch Engine Journal (Matt G. Southern), reporting a newer Ahrefs study. 2 March 2026.
- AI Overview citations now 54% from organic rankingsBrightEdge. 18 September 2025.
- GEO: Generative Engine OptimizationAggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan and Deshpande, KDD 2024. Submitted 16 November 2023, revised 28 June 2024.
- Optimizing your content for inclusion in AI search answersMicrosoft Advertising (Krishna Madhavan). 8 October 2025.
- Introducing AI Performance in Bing Webmaster Tools (public preview)Bing Webmaster Blog. 10 February 2026.
- Overview of OpenAI crawlersOpenAI. Documentation, read 28 September 2026.
- Perplexity crawlersPerplexity. Documentation, read 28 September 2026.
- Aleyda Solis: AI search crawlability and why your site's technical foundations decide your visibilityHumans of Martech (podcast). 13 January 2026.
- 12 experts on AI search strategy in 2026Peec AI. 2026.
How I checked this. I opened every source linked here and checked each figure and each quotation against the source page, most recently on 3 October 2026. Anything in a box marked "My reading" is my own interpretation, not a finding of the studies. Studies change and new ones appear, so I will update this piece and change its dates when the evidence does.