Basics

AI Search Visibility: How Assistants Choose Brands

How ChatGPT, Gemini and Google AI Overviews decide which brands to name and which sources to cite, and what actually moves AI search visibility for a brand.

AI search visibility is how often AI assistants name your brand, and cite your pages, when people ask questions about your category. ChatGPT, Gemini and Google AI Overviews build their answers from two things: what the model learned in training and the web pages it retrieves at the moment of the question. Brands that are described clearly and often across the sites those systems retrieve get named; brands that only describe themselves on their own site usually do not.

This article explains how each engine assembles an answer, what that means for which brands and sources make it in, and which kinds of work tend to move the result. For the broader picture, see what AI visibility is.

How do AI assistants build an answer?

All three engines covered here combine a language model with web retrieval. The rough sequence is:

  1. Interpret the question. The system works out what the user wants and, often, breaks it into several sub-questions.
  2. Retrieve. It runs searches and pulls in pages that look relevant.
  3. Write. The model writes an answer using the retrieved text plus what it already knows, and attaches some of the pages as sources.

The model’s training gives it a prior sense of which brands belong to a category. Retrieval updates and narrows that with current pages. A brand needs to be present in both to be named reliably: known to the model, and described in the pages that get retrieved for that kind of question.

How Google AI Overviews choose sources

Google’s documentation is explicit that there is no special optimisation needed for AI Overviews or AI Mode, and no extra technical requirements. A page has to be indexed and eligible to show a snippet in normal Google Search. Google also says no new machine-readable files or special schema are required.

What Google does describe is query fan-out: AI Overviews and AI Mode may run several related searches across subtopics and data sources to build one response. That has a practical consequence. The pages cited under an overview are not only the ones ranking for the exact query the user typed; they can come from searches on adjacent questions. A detailed page answering “how to invoice a client in another currency” can end up cited for a broader question about freelance accounting tools.

Not every search shows an overview. In Pew Research’s March 2025 browsing data, 18% of Google searches produced an AI summary, rising to 53% for searches of ten words or more. Longer, question-shaped searches are where overviews show up most, which is also where buyer research happens.

How ChatGPT chooses brands and sources

When ChatGPT answers with web search, the pages come from OpenAI’s search systems. OpenAI’s crawler documentation draws three lines:

  • OAI-SearchBot surfaces websites in ChatGPT’s search features. OpenAI states that sites that opt out of OAI-SearchBot will not be shown in ChatGPT search answers, apart from navigational links.
  • GPTBot collects content that may be used to train models. Blocking it signals that your content should not be used for training.
  • ChatGPT-User fetches a page when a user’s action triggers it, not as an automatic crawl.

These are separate decisions. A site can block training (GPTBot) and still allow search (OAI-SearchBot). A blanket disallow rule or a CDN bot setting can block both without anyone deciding to, which removes the site from ChatGPT search answers. It is the first thing to check.

Beyond access, ChatGPT recommendations reflect what the retrieved pages say. If the top comparison articles, review pages and forum threads for a category list four brands, those four are the likely answer.

How Gemini chooses brands and sources

Gemini is Google’s assistant, and Google documents that Gemini models can ground their answers in Google Search results. In practice that means the same fundamentals that make a page retrievable by Google help it reach Gemini answers.

Why do the answers keep changing?

Because generation involves randomness and retrieval shifts with the index, the same question asked twice can return a different list. A 2025 study by SparkToro and Gumshoe, with volunteers running 12 prompts 2,961 times across ChatGPT, Claude and Google’s AI, found less than a 1 in 100 chance that ChatGPT or Google’s AI would give the same list of brands in two responses. Order varied even more. The authors argue that how often a brand appears across many runs is meaningful, while its rank in a single answer is not.

For measurement this means two things. Ask each question more than once, and report visibility as a rate across many answers, not a “position” from one screenshot. We cover the method in how to run an AI visibility audit.

What actually moves AI search visibility?

There is less hard evidence here than the volume of advice suggests. What follows sticks to what is documented or directly observable in audits.

1. Let the right crawlers in

Check robots.txt, firewall and CDN bot rules for OAI-SearchBot and Googlebot. If the search crawlers cannot fetch the site, nothing else matters. This takes minutes and is the most common fixable cause of a brand being absent from cited sources.

2. Be present where the assistants look

When you audit the citations behind answers in a category, review sites, comparison articles, forum threads, YouTube and reference sites often appear next to the brands’ own pages. If a competitor is named and you are not, look at which third-party pages were cited in that answer. Those are the pages that shape the answer, and the practical work is earning a place on them: getting included in comparison roundups, collecting reviews on the sites that get cited, answering real questions in communities where you have a genuine reason to be there.

3. Write pages that answer specific questions

Because of query fan-out and the way assistants pull passages, specific pages do better than broad ones. A page that answers one buyer question directly in its first paragraph, then supports it, is easier to cite than a general product page. Cover the comparison, pricing, use-case and how-to questions buyers actually ask, not only “best X” lists.

4. Add facts, sources and quotations

The research that coined the term generative engine optimization tested this. In Aggarwal et al.’s GEO study, adding citations to sources, quotations and statistics to content were among the most effective changes, improving visibility in generated answers by up to about 40% in their benchmark. Keyword stuffing did little and in one test made things worse. The study used its own test engines, so treat the numbers as directional, but the direction matches what you see in cited pages: specific, checkable claims get used.

5. Keep the brand description consistent

Assistants match a brand to a category through repeated descriptions across sites. If your homepage, directory listings, review profiles and press mentions describe the product differently, the association is weaker. Say what the product is, who it is for and how it differs, in the same words, everywhere you control.

6. Fix what is said, not only whether you are named

A mention can be negative. If assistants repeatedly describe the brand as expensive or hard to set up, that comes from somewhere: reviews, forum threads, an outdated comparison. Find the sources and address the substance.

What does not move it much

  • Special files or markup aimed at AI. Google states no special schema or AI text files are required for AI Overviews. Structured data is still useful for normal search features.
  • Repeating keywords. The GEO study found keyword stuffing ineffective.
  • Chasing one answer. Because answers vary run to run, rewriting a page because one screenshot left you out is reacting to noise. Look at rates across many runs.

How AI search optimization fits with SEO

AI search optimization builds on SEO. Google ties AI Overview eligibility directly to normal indexing, and ChatGPT search depends on its own crawl of the open web. The differences are in emphasis: more weight on third-party coverage, on specific answer-first passages, and on measuring brand mentions rather than rankings. For how the labels relate, see GEO vs AEO vs SEO.

How to measure AI search visibility

Pick 20 to 30 buyer questions that do not name the brand, ask each on ChatGPT, Gemini and Google more than once, and record mention rate per engine, share of voice against competitors, and the sources cited. Repeat with the same questions later to see whether the work moved anything.

AI Visibility Desk does this for agencies: it asks each question on ChatGPT, Gemini, Google AI Overviews, Google AI Mode, Perplexity and Claude (twice on ChatGPT and Gemini, because their answers vary), scores mentions, position, share of voice and citations, checks the site’s crawler access, and publishes a report under the agency’s brand. Audits can re-run weekly or monthly, so the report shows the trend. See the plans if you want to try it on a client.

Frequently asked questions

What is AI search visibility?

AI search visibility is how often a brand is named, recommended or cited when people ask AI assistants such as ChatGPT, Gemini or Google AI Overviews about its category. It is usually measured as the share of relevant answers that mention the brand, alongside which competitors and which sources appear.

How does ChatGPT decide which brands to recommend?

When ChatGPT searches the web, it builds the answer from pages it retrieves and from what the model learned in training. Brands that appear often and favourably in the pages it finds, such as reviews, comparisons and forum threads, are more likely to be named. Sites that block OAI-SearchBot are not shown in ChatGPT search answers.

Do I need special markup to appear in Google AI Overviews?

No. Google says there are no additional requirements and no special optimisation needed for AI Overviews or AI Mode. A page must be indexed and eligible to show a snippet in normal search. Good SEO fundamentals and content that answers the question clearly remain the route in.

How is AI search optimization different from SEO?

SEO aims to rank a page in a list of links. AI search optimization aims to have the brand named in the written answer and its pages cited as sources. Much of the groundwork is shared, but AI answers lean more on third-party mentions, clear factual passages and coverage across many sites.

Written by the AI Visibility Desk team

We build AI visibility audits for agencies and run them across ChatGPT, Gemini, Google AI Overviews, AI Mode, Perplexity and Claude every day. Guides are checked against what the audits show and updated when the engines change. About us.

Run your first audit free

Enter a client’s website, let the AI fill in competitors and buyer questions, and get a branded report across six AI engines. No card needed.

Start free