How do ChatGPT, Perplexity and Google AI choose which sources to cite?
They search first, then write. ChatGPT search, Perplexity and Google's AI Overviews run live web searches, pull a set of pages, and cite the ones that answer the question most directly. A brand gets cited when its pages can be crawled, are already in the index, answer the exact question in plain language, and are backed up by other sites.
Do AI assistants search the web or answer from memory?
Both, depending on the question. A model answers some questions from what it learned in training. For anything current, specific or commercial, the major assistants now run a live search and write the answer from the pages they retrieve. Those retrieved pages are the ones that get cited and linked.
That retrieval step is where a brand can compete. Training data is fixed until the next model. Search results are not.
How does ChatGPT find pages to cite?
ChatGPT search relies on OpenAI's crawler, OAI-SearchBot. OpenAI's documentation says sites that block OAI-SearchBot in robots.txt will not appear in ChatGPT search answers, and that robots.txt changes take about 24 hours to take effect. Blocking GPTBot, the training crawler, is a separate setting and does not remove a site from search.
The practical check is simple: open your robots.txt and confirm OAI-SearchBot is not disallowed. Plenty of sites block it by accident through a blanket bot rule or a security plugin.
How does Google choose sources for AI Overviews and AI Mode?
Google says a page must be indexed and eligible to show in Search with a snippet to appear as a supporting link in AI Overviews or AI Mode. It also says there is no special markup required, and that normal SEO fundamentals still apply.
Google describes a technique called query fan-out: the system runs several related searches across subtopics before writing a response. That means one buyer question can pull sources from many narrower queries, which rewards a site that covers a topic in depth instead of one page per broad keyword.
What makes one page more citable than another?
In our work on owned and client sites, the pages that get cited share four traits. We call them the Answer Layer.
Readable: the crawler can fetch the page and the content is in the HTML, not hidden behind scripts. Resolvable: the brand is a clear entity, with consistent names, schema and profiles that point to each other. Citable: the page answers a specific question in its first lines, with numbers and sources a model can quote. Referenced: other credible sites mention the brand, so the model has independent confirmation.
Most brands fail on citable. Their pages describe the company instead of answering the buyer's question.
Can you pay to be cited?
Not in the organic answer. Ads are appearing alongside some AI answers, but citations inside the answer come from retrieval and ranking. The only way in is to be the most useful source the system can find.
Common follow-up questions
Does schema markup get my brand cited?
Not by itself. Google says no special structured data is required for its AI features. Schema helps a brand resolve as a clear entity, which supports citation, but a page still has to answer the question.
How long does it take to start getting cited?
It varies by category and competition. On CalmGut, a new brand with almost no backlinks, AI-feature impressions began in late June 2026, while the site rebuild was underway. Earning clicks and a top position took longer and is still in progress.
Does llms.txt help?
It is a low-cost signal that gives AI systems a plain summary of your site. No major engine has confirmed it as a ranking factor, so treat it as useful housekeeping, not a strategy.
Sources
- OpenAI, Overview of OpenAI crawlers. https://developers.openai.com/docs/gptbot
- Google Search Central, AI features and your website. https://developers.google.com/search/docs/appearance/ai-features
- Abovance, CalmGut case study. https://abovance.com/case-studies/calmgut/
Want to know where your brand stands?
The AI visibility audit shows which questions your buyers ask, who AI cites instead of you, and what to fix first. Complimentary for qualified brands.