Plenty of brands have a nagging feeling that AI assistants are steering buyers towards their competitors, but no clean way to confirm it. The reassuring part is that you can check by hand, with nothing but a browser and a spreadsheet. This is the manual method we run before we automate anything, and it settles the question most marketers are really asking: how do I know if competitors are being mentioned in AI responses instead of us?
One thing to keep in mind before you start. When you ask an assistant a question, it usually does not search the exact words you typed. It rewrites your question into its own searches and answers from what those return, which is part of why how ChatGPT chooses its sources looks nothing like a Google ranking. That has a direct effect on how you should test.
Step 1: Ask the questions your buyers ask, not your brand name
Searching an assistant for your own company name tells you very little. Of course it can describe you once you have named yourself. The real test is the question a buyer asks when they do not yet know who you are, and when a competitor could just as easily be the answer.
Write down five to ten of those questions. The useful shapes are the category question ("what is the best [category] for [use case]?"), the shortlist question ("who should we use for [need]?"), and the comparison question ("what are the top options for [problem], and how do they compare?"). Phrase them the way a customer speaks, not the way your marketing team writes. If you sell multi-currency accounting software to small firms, ask "what is the best accounting tool for a small business that invoices in several currencies?" rather than anything with your product name in it.
Step 2: Run every question across all the AI platforms
A brand can be the default answer on one assistant and completely absent on another, so testing a single one is close to useless. Run each of your questions through all six of the surfaces buyers actually use:
- ChatGPT
- Claude
- Gemini
- Perplexity
- Google AI Overviews (the answer box at the top of a normal Google search)
- Google AI Mode
They do not behave the same way, which is the whole reason to check each one. Gemini tends to list a lot of brands in a single answer. Google AI Overviews is short and rarely names a clear winner. ChatGPT and Perplexity will often browse the live web and cite sources as they go, while Claude leans on what it already knows rather than browsing by default. We cover those differences in full in what AI search visibility is and how it is measured. For now the point is simple. This is how you tell whether your brand appears in Perplexity, Gemini, Claude, and Google AI answers: by looking at each of them, rather than assuming they agree.
Step 3: Run each question more than once
This is the step most people skip, and it is the one that changes the result. Ask the same assistant the same question twice and you will often get two different lists. The model rewrites your question into fresh searches each time, those searches pull slightly different sources, and so the set of brands it names moves from run to run. A single run is a snapshot, and a lucky or unlucky one at that.
Run each question at least three times per platform, ideally in a fresh chat or an incognito window so past conversation and your account history do not colour the answer. If your brand shows up in one run out of three, treat that as noise rather than visibility. What you are looking for is the pattern that holds across repeated runs.
Step 4: Record whether you were named, recommended, or cited
Being in the answer covers three different things, and they are worth tracking separately.
- Named: the assistant mentions your brand at all, even in a long list.
- Recommended: the assistant actively puts you forward as a good choice, not just one option among ten.
- Cited: the assistant links to a source, and that source is either your site or somewhere that talks about you.
For each run, note which of the three happened, for you and for every competitor that appears. A simple grid does the job: platforms down the side, your brand and your main rivals across the top, and a mark in each cell for named, recommended or cited. Pay attention to where you land in the answer too, since a brand mentioned in the first line carries far more weight than one buried at the bottom that most readers never reach.
Step 5: Read who owns the answer
Now tally the grid, and the picture usually resolves fast. Count how often each brand was named and recommended across every platform and every run. The brand that appears most consistently, and earns the clearest recommendations, is the one that currently owns the category answer. If that is a competitor and not you, you have your confirmation, and it is worth knowing which competitor and on which platforms.
Then look at the sources. When an assistant cites its reasoning, the pages it leans on are usually not brand websites at all. They tend to be reviews, directories, forums and publisher round-ups, the third-party places these tools trust for a category. If your competitors keep turning up in those cited sources and you do not, that gap is most of the reason they are in the answer and you are missing. We go through why that happens, and what to do about it, in why brands do not show up in AI answers.
When to use a structured check instead
The manual method is honest and it is free, and for a first look it is often enough. Its limits show up when you try to make it repeatable. Five questions across six platforms three times each is ninety runs, and doing that every month by hand, scoring it consistently, and comparing like for like over time is more than most teams will keep up.
That is the point at which a structured check earns its place. Our free AI visibility check runs the same exercise across ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews and Google AI Mode, repeats each query to smooth out the run-to-run variance, and returns a single score out of 100 with a platform-by-platform breakdown and the competitors named in your place. It is the automated version of the afternoon you would otherwise spend with a spreadsheet, and it gives you a baseline you can measure against next month.
Frequently asked questions
How do I know if competitors are being mentioned in AI responses instead of us?
Ask the assistants the questions your buyers ask, such as the best option in your category or who they should use for a specific need, and run each one across ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews and Google AI Mode. Record which brands get named and recommended in each answer, and repeat every question several times because the results vary from run to run. If competitors appear consistently and you do not, you have your answer. A free AI visibility check runs the same exercise and scores it for you.
How can I tell whether my brand appears in Perplexity, Gemini, Claude, and Google AI answers?
Check each platform separately, because a brand can be visible on one and absent on another. Ask each of Perplexity, Gemini, Claude, Google AI Overviews and Google AI Mode a category question a buyer would ask, without using your brand name, and note whether you are named at all. They behave differently: Gemini tends to list many brands, Google AI Overviews rarely names a clear winner, and Perplexity cites its sources inline. Testing all of them is the only reliable way to know where you stand.
How can I tell whether my brand is showing up in AI answers?
Run the buyer-shaped questions in your category through each major assistant several times, then count how often your brand is named versus your competitors. Showing up once in one run tells you little on its own, so look for the pattern that holds across repeated runs and across platforms. For a scored, repeatable version of the same check, a free AI visibility check does it across every platform and returns a result out of 100.
What should I do if competitors are getting AI mentions and we are not?
First, find out which sources the assistants cite when they name your competitors, because those are usually third-party pages such as reviews, directories, forums and publisher round-ups rather than brand websites. Being present and well regarded on the sources AI trusts for your category is most of the work of getting into the answer. Build that third-party footprint, keep your entity details consistent across the web, and publish content that answers the exact questions buyers ask.
Why isn't one run enough to check AI answers?
Because assistants rewrite your question into their own searches each time you ask, and those searches pull slightly different sources, the set of brands named moves from one run to the next. A single run is a snapshot that can flatter or mislead you. Run each question at least three times per platform, ideally in a fresh chat or incognito window, and trust the pattern that repeats rather than any single answer.