You do not need a $200-a-month tracking dashboard to find out whether AI tools recommend your business. You need four prompts, five minutes, and a plain way to read what comes back. Here is exactly how to run the test yourself, today, with no software to install and nothing to sign up for.
This matters more than it might feel like from behind a keyboard. Somewhere right now, a person with money to spend on exactly what you offer is typing a question into ChatGPT instead of Google, and the name that comes back is either yours or it is not. There is no ad you can buy to force your way into that answer, and no amount of traditional SEO alone guarantees it either. The only way to know where you actually stand is to ask the same question the AI is being asked, from the outside, the way a real customer would.
Run it once now to see where you stand, then run it again every month or two. AI models retrain, competitors improve their own structured data, and an answer that named you in March can quietly stop naming you by June with no notification of any kind. There is no dashboard that pings you when that happens. Checking by hand, on a schedule, is currently the only reliable way to know.
How to check if ChatGPT and Perplexity recommend you
- Open ChatGPT and ask: “Who is the best [your service] in [your city]?” Do not mention your business name, since the goal is to see what it volunteers unprompted.
- Ask the same question in Perplexity, and note whether it shows source citations, and whose site those citations point to. Perplexity is more transparent about its sources than most, which makes it the easiest of the four to actually learn from.
- Search the same phrase in Google and check whether an AI Overview appears above the normal results, and who it names. Note that the AI Overview and the ten blue links below it can name different businesses entirely.
- Ask Claude a slightly more specific version: “I need [your specific service], who should I call in [your city]?”
- Write down, for each of the four, whether you were named, mentioned in passing, or absent entirely, and repeat the exact same prompts next time so the comparison is apples to apples.
Try a second round with slightly different phrasing, too: “recommend a [service]” gets a different answer than “who is the best [service]” often enough to be worth checking both. AI models are sensitive to how a question is asked in ways search engines mostly are not, so a single prompt is a data point, not a verdict, and drawing a conclusion from one phrasing alone is how businesses either overreact to a bad result or take false comfort from a good one.
If you would rather not run this by hand, an AI visibility audit is the structured version of exactly this: twenty fixed prompts across three platforms, with the raw answers recorded so you have a baseline to measure against later.
How to read the answer: cited, mentioned, or ignored?
Not every appearance in an AI answer is equal, and treating all three of these the same hides real progress or a real gap.
| What you see | What it means |
|---|---|
| Cited: your business named directly, often with a source link | The AI model trusts your site enough to treat it as a source. This is the goal. |
| Mentioned: you appear in a list alongside competitors, no elaboration | You exist in the model’s training or retrieval data, but nothing about your site stood out enough to lead with |
| Ignored: a confident answer is given, and it is not you | A competitor’s structured data, reviews, or content is currently doing the job yours should be doing |
Being merely mentioned instead of cited is a more common outcome than people expect, and it is worth taking seriously rather than treating as a consolation prize. A customer who sees your name buried in a list of six alongside competitors is far less likely to choose you than one who sees your name given as a direct, confident recommendation. The gap between mentioned and cited is where most of the real business value in this whole exercise actually lives, and it is the gap the three fixes below are specifically aimed at closing.
Why ChatGPT isn’t recommending you yet
- Your website has no schema markup telling AI systems what you actually do, so a model has to infer it from unstructured text instead of reading it directly
- Your content answers questions vaguely instead of directly and extractably, building up to a point over several paragraphs instead of stating it first
- Your reviews and mentions across the web are thin, inconsistent, or outdated, giving an AI model little independent confirmation of what you actually do well
- A competitor’s site is more explicitly structured around the exact question people are asking, even if your actual service is comparable or better
- Your Google Business Profile and website disagree on basic facts, like hours or service area, which erodes AI confidence in either source being reliable
The three fixes that move you into the answer
- Add real structured data. Organization, LocalBusiness, Service, or Article schema, matched to what your page is actually about, gives AI models something concrete to extract instead of guessing. This is usually the single highest-leverage fix on this list, and often the one most sites skip entirely.
- Write direct-answer content. State the answer to a likely question in the first sentence of a section, then support it with the reasoning and detail afterward. AI systems extract the clean, front-loaded answer, not the one buried three paragraphs into scene-setting, so a page that makes a human reader wait for the point makes a machine give up on it entirely.
- Make your facts agree everywhere. Your website, Google Business Profile, and any directory listings should state your hours, services, and location identically. Disagreement between sources is exactly what makes an AI model less confident in citing any of them, since it has no way to know which version is current.
Of the three, structured data is usually the fastest to implement and the easiest to verify, since you can check it with a free schema validator immediately after adding it. Direct-answer content takes longer, because it means rewriting existing pages rather than adding a tag, and cross-source consistency takes an audit across every place your business is listed, which for most businesses is more places than they remember signing up for. Start with structured data if you only have time for one this month. It is the highest ratio of impact to effort of the three, and it rarely requires touching the words on the page at all.
If your business is specifically local and service-based, our restaurant-focused walkthrough of appearing in Google AI Overviews applies the same three fixes to a very concrete example, and our llms.txt explainer covers one more piece of the structured-data picture. For the bigger-picture reasoning behind why a site that ranks fine can still get skipped by AI entirely, that gap is exactly what our AI Search Readiness & AEO service is built to close.
One last thing worth expecting rather than being surprised by: fixing the three items above does not guarantee an appearance on your very next check. AI models update on their own schedules, and a change you make today might not be reflected in an answer for weeks. That lag is real and it is normal. It is not a sign the fix did not work, and the businesses that give up after one unchanged result before the model has had time to catch up are the ones who conclude, incorrectly, that none of this matters.