What AI Overviews and ChatGPT say about your business, and how to check

A customer who used to read five reviews can now read one paragraph. Search a business name with “reviews” and Google often answers with an AI Overview above the results. Ask ChatGPT or Gemini “is this company reputable?” and you get a confident summary of praise and complaints.

That paragraph is now part of your reputation. You did not write it, and you cannot edit it, but you can read it, understand where it comes from and work on the sources. This guide shows how, and why you should never judge it by a single answer.

Where AI answers about you appear

• Google AI Overviews sit at the top of many search results, including searches for a business name with “reviews”, “legit” or “complaints”. They link to sources.
• Google AI Mode is a separate, conversational search mode; Google’s documentation for site owners covers it together with AI Overviews.
• ChatGPT, Gemini and other assistants answer either from what the model learned in training or after searching the web, depending on the product and the mode. The two can give very different answers about a small business.

Google’s documentation for site owners explains that its AI features use a technique called query fan-out: “issuing multiple related searches across subtopics and data sources”. In practice an AI Overview about your business can lean on pages that do not appear in the first ten results for your name.

One answer proves nothing

Before you check, accept one thing: AI answers are not stable. Ask the same question twice and you may get different facts, a different tone and different sources.

SparkToro, with Gumshoe.ai, measured this in January 2026. 600 volunteers ran 12 prompts through ChatGPT, Claude and Google’s AI, 2,961 times in total. The headline finding: there is “a <1 in 100 chance that ChatGPT or Google’s AI, if asked 100X, will give you the same list of brands in any two responses”, and about 1 in 1,000 that two lists come in the same order. The authors dismissed ranking positions in AI answers as meaningless, and concluded that “visibility % across dozens to hundreds of prompts run multiple times is a reasonable metric”.

For reputation the lesson is the same. Do not screenshot one answer and panic or celebrate. Ask several times and count.

“any tool that gives a ‘ranking position in AI’ is full of baloney”

Rand Fishkin, SparkToro, 27 January 2026

How to check it yourself in 15 minutes

1. Google, private window. Search your name reviews and is your name legit from the country where your customers are. If an AI Overview appears, copy its text and open every source it links to.
2. Ask an assistant the buyer’s question. Something like: “I am a customer in [country] thinking about using [your name]. What do you know about this business? Is it reputable? What do people praise and what do they complain about?”
3. Ask it three times, in new chats. Note in how many runs it recognised the right business, whether it confused you with a namesake, and the tone: positive, neutral or negative.
4. Repeat in a second assistant. ChatGPT and Gemini can disagree about the same business, as in the example below.
5. Write down every specific claim. “Complaints about delayed refunds”, “closed in 2023”, “founded in 1975”. Each claim either has a source on the web or is a mistake.

What this looks like on a real business

Our public sample report is a real, unedited check of Joe’s Pizza in the United States, run on 2 October 2026. The same customer question went to two models, three runs each, without web search. Gemini recognised the business in 3 of 3 runs, with a positive tone. ChatGPT recognised it in only 1 of 3 runs, with a neutral tone overall. The Google AI Overview for the same name described a classic New York slice shop and summarised both praise and common complaints.

Same business, same day, three quite different pictures. That is normal, and it is why we report presence and tone across runs rather than one quote.

What shapes the answer

AI answers are built from what the web already says about you, and for its own features Google says there is nothing special to add: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” Site owners are told to apply the same SEO basics as for Search, and that they do not need “new machine readable files, AI text files, or markup”.

One study points at what matters most for brands. Ahrefs looked at 75,000 brands in 2025 and found that branded web mentions had the strongest correlation with brand visibility in AI Overviews, 0.664, against 0.218 for backlinks. The author is careful to add that “correlation ≠ causation”. Still, it points one way: how often and how people write about you around the web is what AI tends to repeat. We look at that study more closely in E-E-A-T and brand mentions: what the evidence says.

What you can realistically change

What you should not do is write fake reviews or fake “articles” to feed the models. Google’s review policy and the FTC rule on reviews in the US both ban fake reviews, and an answer built on obviously planted content can turn against you.

Keep an eye on it

AI Overviews change with the results underneath them; model answers change more slowly. A sensible rhythm is to look at the Overview often and at the assistants monthly. WhatsMyRep reads the Google AI Overview for your name every day, asks ChatGPT and Gemini the buyer’s question every week, three runs each, and alerts you when an answer turns negative. More on the AI visibility page.