If a potential customer asks ChatGPT to recommend a business like yours, will they see the same companies if they ask Gemini? And if they repeat the question tomorrow, will the answer still be the same? New research suggests businesses should not assume so.
A September 2026 study examined 1,536 responses to commercial product questions across ChatGPT, Gemini, and Google AI Overviews. It found that recommendations could change when questions were repeated, while the sources used by different AI platforms varied considerably. ChatGPT and Gemini shared just 5.4% of displayed source domains on average for the same questions, with no source domains in common in 76.7% of the comparisons.
For businesses, this raises an important point about AI visibility. Appearing in one answer on one platform does not necessarily mean your business has established a strong presence across AI search. Understanding that visibility requires a broader view of where your business appears, which competitors are being recommended and which sources AI platforms are using to reach those answers.
What we Recommend
We recommend treating AI visibility as a broader measurement exercise rather than checking a handful of prompts and assuming the results represent your overall position.
Start by understanding how your business and its competitors appear across the main AI platforms, which sources are influencing those answers and where clear visibility gaps exist.
From there, SEO, authoritative content, digital PR and improvements to how clearly your expertise is communicated online can be prioritised around the areas most likely to make a meaningful difference.
Why can ChatGPT and Gemini give different recommendations?
AI platforms do not work like a traditional Google results page where businesses occupy relatively identifiable ranking positions.
When someone asks for a recommendation, an AI platform generates an answer based on the question, the information it can access and the way its own systems evaluate that information. Different platforms can therefore reach different conclusions.
The September study illustrates just how different those information sources can be. Researchers tested 117 product questions repeatedly across ChatGPT, Gemini, their APIs and Google AI Overviews. Even when ChatGPT and Gemini received the same question, the websites displayed as sources were often different.
This matters because two prospective customers researching the same type of business could encounter quite different recommendations depending on where they search.
Can the same AI platform give a different answer twice?
Yes, and this is one of the more important findings for businesses trying to measure AI visibility.
The study found that recommended products often changed when requests were repeated. Separate September research also found considerable variation when identical questions were run more than once across AI platforms.
Imagine asking ChatGPT, “Which are the best commercial ventilation companies in the UK?” and seeing your company recommended. It would be tempting to record that as a success.
However, running the same question again might produce a different collection of companies. A potential customer asking from another location or phrasing their question differently could also receive another answer.
This is why a single screenshot provides useful evidence of an appearance, but relatively little evidence of consistent visibility.
What should businesses measure instead of a single AI ranking?
Traditional SEO has trained businesses to think about positions. You might rank third for one keyword and seventh for another.
AI search requires a slightly different way of looking at performance because there may be no permanent equivalent of “position three”.
A useful AI visibility review should instead examine a representative group of questions that real customers might ask. Those questions can then be checked across platforms such as ChatGPT, Google Gemini and Google AI search experiences.
The important questions become:
- How frequently is the business mentioned?
- Which competitors appear more regularly?
- Is the business actively recommended or simply referenced?
- Which websites are being cited when recommendations are made?
- Does the picture remain reasonably consistent when checks are repeated?
That produces a much more useful benchmark than asking one question once.
Why do the sources behind AI answers matter?
The websites being used to support AI answers can tell businesses almost as much as the recommendation itself.
An AI platform may rely on a company’s own website for information about its expertise, services and products. For other questions, it might draw upon industry publications, review websites, news coverage, directories, comparison websites or other independent sources.
Research published in September reinforces this point. One large study of different AI engines found relatively little overlap between the websites cited by different platforms, including only 8.4% overlap between ChatGPT and Google AI Overviews in its dataset.
For a business, analysing those sources can uncover practical gaps. A competitor might be repeatedly mentioned because it has clearer specialist content, stronger third-party coverage or a better-established presence on websites that AI systems frequently reference.
That gives the business something tangible to investigate rather than simply trying to “rank in ChatGPT”.
Where do SEO and content fit into AI visibility?
The growing industry term for improving visibility within AI generated answers is Generative Engine Optimisation (GEO). In practice, however, much of the work overlaps with established digital marketing disciplines.
A business still needs a website that clearly explains what it does, who it helps and why its expertise is credible. Service pages need enough detail to answer genuine buying questions. Articles and guides can address the more specific questions customers ask during research. Case studies, original insight and first hand experience can provide evidence that generic content cannot.
Visibility beyond the company’s own website matters too.
Digital PR, relevant industry coverage, trusted directories, reviews and other credible third-party references can strengthen the information available about a business across the wider web.
This is where a cohesive approach becomes valuable. SEO, content and digital PR can each contribute different pieces of the information AI systems encounter when researching a topic or business.
How can businesses establish their current AI visibility?
The sensible starting point is to establish a benchmark. An SEO & AI Visibility Audit can examine the commercial questions customers are likely to ask across multiple AI platforms, rather than checking the company name directly. After all, someone who already knows your name is at a different stage from someone asking an AI assistant which businesses they should consider.
The audit can record whether the business appears, which competitors appear alongside or instead of it, how the company is described and which sources support the response.
Repeating those checks also helps separate occasional appearances from stronger patterns.
From there, businesses can identify where improvement is realistically needed. That might involve strengthening important service pages, creating content around unanswered customer questions, improving evidence of expertise or building greater authority through relevant third party coverage.
Why is ongoing monitoring likely to become more useful?
AI search is moving quickly. Models change, sources change and the answers generated for the same question can change too.
That makes a one off audit useful as a benchmark, while ongoing monitoring can show whether the wider picture is improving.
For example, a business might initially appear in two out of ten relevant questions on ChatGPT and rarely appear on Gemini. Several months later, repeated checks might show broader coverage, more frequent recommendations or new third-party sources supporting those mentions.
That is a more meaningful way to assess progress than expecting an AI recommendation to remain fixed indefinitely.
FAQs
Does appearing in ChatGPT mean my business will appear in Gemini?
No. Different AI platforms can use different sources and produce different recommendations, even when answering the same question.
Why does an AI recommendation sometimes change?
AI-generated answers can vary between requests because they are generated dynamically. The information retrieved and the way the platform constructs its response can also differ between searches.
What is an AI Visibility Audit?
An AI Visibility Audit checks how frequently a business appears for relevant customer questions across several AI platforms. It can also examine competitor mentions and the sources supporting those answers.
What is GEO?
GEO stands for Generative Engine Optimisation. It describes work intended to improve how a business, brand or website is represented and discovered within AI-generated answers.
Can SEO help improve AI visibility?
SEO can form an important part of the process because clear website structure, useful content and demonstrable expertise make information about a business easier to find and understand. Wider signals such as authoritative third-party coverage and digital PR can also contribute to the information available to AI platforms.