How to Check if AI Models Cite Your Brand

Run a fixed set of real buyer prompts through ChatGPT, Google AI Overviews and AI Mode, Gemini, Perplexity, and Copilot, with personalization turned off, several times each. Log whether your brand is mentioned, whether it is linked as a source, and who appears instead. Then confirm it with free first-party data: the AI Assistant channel in GA4, utm_source=chatgpt.com, and the AI Performance report in Bing Webmaster Tools. Report a visibility percentage, not a rank, because AI answers change from run to run.

Why this is worth checking now

AI answers are now a front door to buying decisions. OpenAI reported more than 900 million weekly active users for ChatGPT in February 2026. Google said in July 2026 that AI Mode had passed one billion monthly active users and that the Gemini app had 950 million.

Buyers are using these tools on purpose. A McKinsey survey of US consumers published in October 2025 found that half now intentionally seek out AI-powered search, and 44% of AI search users call it their primary source of insight.

The uncomfortable part for brands: the same McKinsey research found that a brand's own sites make up only 5 to 10% of the sources AI search references. Most of what an AI says about you comes from other people's pages.

The traffic is still small but growing. SE Ranking's study of 101,574 websites found AI platforms sent 0.32% of all website traffic in early 2026, up from 0.24% in 2025, and ChatGPT accounted for 74.78% of AI referral traffic.

Mentions vs citations: know what you are measuring

These two terms get mixed up constantly, and they mean different things.

Track both. A brand that is mentioned but never cited is relying on other sites to tell its story. A brand that is cited but never mentioned is supplying facts without getting credit.

How to check manually, step by step

You can do a solid first check in under an hour with nothing but a spreadsheet.

1. Build a prompt set from real buyer questions

Write 15 to 30 prompts across three types:

• Discovery prompts with no brand name, such as "best project management tools for small agencies."

• Comparison prompts, such as "[your brand] vs [competitor]" or "alternatives to [competitor]."

• Branded prompts, such as "what is [brand]" or "[brand] pricing," to check accuracy and sentiment.

Pull wording from sales calls, support tickets, and your Search Console queries. Discovery prompts matter most, because that is where a buyer has not decided yet.

2. Check every platform separately

Cover ChatGPT with search on, Google AI Overviews, Google AI Mode, Gemini, Perplexity, and Microsoft Copilot. Do not assume one result stands in for another. Ahrefs found that AI Mode and AI Overviews cited the same URLs only 13.7% of the time in its 2025 study, even though both are Google products.

3. Turn off personalization

Your own history skews the answer. In ChatGPT, use Temporary Chat with the Unpersonalized option, which OpenAI says "does not use memory, custom instructions, or plugins." For Google, use a signed-out incognito window. Record your location too, since OpenAI notes ChatGPT search may use your general location, and Google results vary by country.

4. Run each prompt more than once

This is the step most people skip, and it changes the conclusion. SparkToro's January 2026 research found AI tools give the same list of brands less than 1 in 100 times, and the same list in the same order roughly 1 in 1,000 times. Ahrefs found AI Overviews have a 70% chance of changing between observations, with 45.5% of cited sources new each time. Run each prompt at least three to five times, on different days.

5. Log the same fields every time

6. Report a visibility percentage, not a rank

Divide the runs where your brand appeared by total runs, per platform. That is your visibility rate. Do the same for citations. Given how much answers move, a "position 2 in ChatGPT" claim from a single run tells you very little. SparkToro's Rand Fishkin called visibility percentage across many prompts and repeated runs "a reasonable metric," and was blunt about tools that report a ranking position.

If you want a quick baseline before building the full spreadsheet, our free AI visibility checker gives you a starting snapshot.

Confirm it with free first-party data

Manual checks show what AI says. Your own analytics show what it is already sending you.

Look at the landing pages in each report. They tell you exactly which pages AI tools are citing, which is the most useful clue for what to build next.

One more technical check: OpenAI's publisher FAQ says that if you want to appear in ChatGPT search, you should make sure you are not blocking OAI-SearchBot. Our robots.txt generator covers the AI crawler directives.

Tracking at scale: which tools to use

Manual checks stop scaling at around a few dozen prompts. Beyond that, tools fall into three groups.

When comparing tools, ask five questions: which platforms it covers, how many times it runs each prompt, whether you can set location and language, whether it separates mentions from citations, and whether it exports the third-party sources being cited.

What to do when AI does not cite you

The check is only useful if it changes what you do next.

• Study the sources that win. The third-party domains cited in your place are your outreach and PR list. Ahrefs found brand web mentions had the strongest correlation (0.664) with AI Overview brand visibility, well ahead of backlinks (0.218).

• Fix what the AI gets wrong. If branded prompts return outdated pricing or features, update your own pages and the third-party profiles the AI is quoting.

• Fill the gaps on discovery prompts. If competitors appear for "best [category] for [use case]" and you do not, publish content that answers that exact question with specifics.

• Check crawler access. Make sure OAI-SearchBot, Bingbot, Googlebot, and PerplexityBot can reach your pages.

For the full optimization playbook, see our guides on LLM SEO and answer engine optimization.

Common mistakes

• Checking once and calling it a result. A single run is an anecdote. Answers change most of the time.

• Checking while logged in. Memory and history bias the answer toward brands you have already researched.

• Only testing branded prompts. Asking "what is [brand]" proves the model knows you, not that it recommends you.

• Treating Google AI Overviews and AI Mode as one surface. They cite very different URLs.

• Ignoring the citation sources. The pages cited instead of you are the most actionable data in the whole exercise.

• Reporting a "rank" in AI. Report visibility rate and citation rate per platform instead.

Important notes

• This article is for educational purposes and reflects the position as of October 2026. AI platforms, analytics features, and tools in this space change quickly.

• Usage figures come from OpenAI and Google. Research figures come from published studies by McKinsey, SE Ranking, SparkToro, and Ahrefs, whose methods and samples differ. None observes the internal systems of any AI provider.

• Tools are named as examples of each category, not as endorsements, and their features were checked against their own websites in October 2026.