All posts
· The Topic Modeler Team

Does ChatGPT Recommend Your Business? Here's How to Check in 15 Minutes

Ten prompts reveal whether ChatGPT recommends your business. Run the 15-minute audit, then fix your coverage with Topic Modeler.

Does ChatGPT Recommend Your Business? Here's How to Check in 15 Minutes

We work with a high-tech manufacturer that builds extremely precise measurement instruments for electronics production. International footprint, serious engineering pedigree, and a mandate for us: improve the visibility of their U.S. business. One of the first things we did was the check this post teaches. We asked the AI engines about their category, the way a real buyer would, and watched what came back.

They were barely there. For a company of that caliber, the absence was jarring, and it sent us into their website looking for reasons. We found several places where the site was suppressing crawlability across many of its pages, machines politely turned away at the door. The remedy was an easy fix once we saw it. Within 45 days of making those pages crawlable and scannable, their AI mentions were up 300%.

That is the pattern worth internalizing: the check takes 15 minutes, the causes are often mundane, and the upside is large. Here is how to run it yourself.

The 15-minute audit

Open ChatGPT, Perplexity, and Google's AI Mode. You want all three, because they retrieve differently and your results will differ across them. Then ask each one a set of questions a real customer in your market would ask. Ten prompts, adapted to your business:

  1. Best [your category] in [your city]
  2. Who should I hire for [the problem you solve] in [your city]?
  3. Top companies for [your specialty]
  4. [Your category] near me, who do you recommend?
  5. How do I choose a [your category] provider?
  6. Is [your business name] any good?
  7. [Your business name] vs [your main competitor]
  8. Alternatives to [the market leader in your space]
  9. What does [your specialty service] cost in [your region]?
  10. Who are the experts in [your niche topic]?

Swap city terms for industry terms if you sell nationally, the way our instruments client would.

What to record

For each prompt on each engine, note three things: whether you appear at all, what the engine says about you if you do, and who gets named instead of you. That last column is the painful one and the most useful. Keep it in a simple spreadsheet; you will want the baseline in 90 days.

How to read your results

Your scores sort into three tiers.

Absent. You never come up. The machines either cannot see your site or cannot connect you to your own category. Check crawlability first, exactly as our client's story suggests, then look at whether your content actually covers the questions being asked.

Mentioned. You appear in lists but without conviction. The engines know you exist and do not yet know why you are the answer. This is usually a depth and structure problem: your trust signals are thin, and your coverage does not mark you as the specialist.

Recommended. The engine names you with a reason attached. This is the tier that mints customers, and whatever earned it deserves protection and reinforcement.

Whatever tier you land in, remember the good news baked into our client's result: these outcomes move. This is the same machinery we described in SEO and GEO are not rivals, and it responds to deliberate work, sometimes fast.

So how do you fix it?

That is the natural next question, and the honest answer is: it depends on why the machines are passing you over. Missing coverage, scattered structure, blocked crawlers, and invisible entity signals all produce the same empty answer, and each needs a different fix.

Diagnosing which one is your problem is what Topic Modeler does. It maps your site the way AI systems read it, shows you where the coverage and structure fall short of your actual expertise, and turns "we're not in the answers" into a fixable list. Run the 15-minute check today. Then find out exactly why the answers came back the way they did.

Want the full Topic Modeler stack?

Five modules + an Enterprise bundle. Foundation projects + ongoing content tooling for AI-search visibility.