Free Competitor AI Visibility Check: See Which Brands AI Recommends Instead of You

The check answers one question: when a buyer asks an AI engine for a recommendation in your category, who gets named, and which sources does the engine lean on?

That gives you three useful outputs:

• A mention log. Which brands appear, on which engine, for which question.
• A cited-domain list. The review sites, directories, forums and publications the engine pulled from when it searched the web.
• A gap list. Questions where like-sized competitors appear and you don't.

It does not give you a position or a rank. Google exposes impressions only for its AI features, so there is no position data to check a rank against. Our guide on why AI has no rank to check covers this. What you get is presence or absence, counted across a sample of answers.

You need a spreadsheet and about an hour.

1. Write 10 to 20 unbranded buyer questions. Phrase them the way a buyer would: "best payroll software for a 20-person agency" or "which CRM works for a small dental clinic in Austin" (hypothetical examples). Never include your brand or a competitor's.
2. Pick your comparison set. Choose three to five competitors close to your size, price and market. Add any category giants in a separate column so they don't distort your read.
3. Remove personalisation. Use a private or logged-out window where you can. If you must stay logged in to ChatGPT, turn memory off in its settings. OpenAI says ChatGPT may use saved memories when it rewrites a search query.
4. Ask each question on each engine. Cover ChatGPT, Perplexity, Gemini, Claude and Google. In Google, note AI Overviews and AI Mode separately: Google says they may use different models and techniques. Our guide to comparing AI Mode and AI Overview answers shows how.
5. Log every answer. Use the columns below.
6. Repeat the run two or three times. Answers vary between runs, so one pass is too thin to act on.

For the counting formulas, see the spreadsheet method for AI share of voice.

Start with mention rate per engine. Divide the answers that name a brand by the answers you logged for that engine. Do it for you and each like-sized competitor. Keep engines separate. In Otterly's 2026 experiment, Microsoft Copilot produced 69% of one test site's tracked AI citations while ChatGPT, Gemini and Google's AI features barely cited it. A blended number would have hidden that.

Ignore the giants when you judge your gap. Citations concentrate heavily on big names. One study of AI search (arXiv 2507.05301, data from March to May 2025) found the top 20 news sources took 67.3% of all citations for OpenAI models. Losing to a household name tells you little. Losing to a rival your size tells you something.

Then sort your cited-domain column by frequency. These are the sites the engines already trust for your questions. When engines cite other sites instead of yours, being named on those sites is the route into the answer. The evidence shows an association, not a guaranteed effect. Practical next moves:

• Check whether you're listed on the top cited review sites and directories, and whether your facts there are right.
• Note which cited articles name your competitors but not you.
• Flag any answer that gets a competitor's facts wrong or names a brand that no longer exists. Our page on AI recommending a defunct competitor explains why that happens.

If like-sized competitors show up and you're absent everywhere, check access before anything else. This is the part you control.

• Your robots.txt rules work per bot. OpenAI documents that OAI-SearchBot controls ChatGPT search, while GPTBot only governs training use. Each setting is independent.
• A firewall or CDN can block an AI crawler that robots.txt allows. OpenAI's help page says eligibility needs both the crawler allowed and your host or CDN letting its published IP addresses through.
• Content that only appears after JavaScript runs may be invisible to AI crawlers. Put prices and core product facts in the HTML.
• Look for accidental noindex, nosnippet or data-nosnippet on key pages.

We cover the rest in why ChatGPT recommends competitors but not you.

Treat the result as a sample, not a census.

• You can't see the real queries. OpenAI says ChatGPT search typically rewrites your question into one or more targeted queries. Your wording is a proxy for an input nobody outside the engine observes.
• Search doesn't always run. When an answer shows no sources, it likely came from what the model learned in training, which you can't change this quarter.
• Cited is not the same as used. Research on how language models cite (arXiv 2606.28358) shows citation output can be changed without changing the sources, so citation counts are a noisy proxy for influence.
• It goes stale. Engines change models and defaults often. Date every run and re-check monthly.

If you report a share figure, read the definition and limits of AI share of voice first, and don't present it as a market share.

The hand method works, but logging dozens of answers takes time. The free checker on this site runs up to three buyer questions (yours, or ones written from your homepage) and scores who gets named. It's a one-off snapshot, not ongoing tracking.

For a quick repeat check, 99Visibility's free AI Visibility Checker does the same job. Use either one to spot the gap, then use your spreadsheet to dig into cited domains and wrong facts.