5 prompts to test whether AI recommends your service business

The fastest way to understand AI visibility is to run the experiment yourself. Open ChatGPT, Gemini, or Perplexity, ideally all three, and paste in the prompts below. Replace the bracketed parts with your trade and your town.

This takes fifteen minutes. What you learn will tell you more than any audit tool.

The ground rules before you start.

Use a fresh or logged-out session if you can. AI tools personalize answers based on your history. You want to see what a stranger sees, not what the tool has learned about you.

Run each prompt twice, in two different tools. The tools disagree with each other constantly. If you’re named in one and invisible in another, that’s normal, and useful.

Write down exactly what happens. Which businesses get named. What the AI says about them. Whether your business appears at all. This is your baseline; you’ll repeat the test monthly.

Prompt 1: The direct ask.

“I’m looking for a [plumber] in [Lincoln, Nebraska]. Who would you recommend, and why?”

This is the money prompt, it’s the closest to what a real customer types. Are you named? If yes, read the “why” carefully: the reasons the AI gives are the evidence it found about you, your reviews, your service descriptions, your profile. That’s your current reputation, as the machines see it. Who gets named instead? Click through to their websites and notice what they have that you don’t: more specific service pages, more reviews, a completed profile. The gap is usually visible in sixty seconds. Does the AI hedge? Phrases like “I don’t have real-time information” mean it couldn’t find enough confident material to make a real recommendation. That’s a content gap, not a penalty.

Prompt 2: The comparison.

“Compare the top [house cleaning] options in [Lincoln, Nebraska], what does each one do well?”

This prompt tests whether the AI has differentiated knowledge about your market. A generic answer (“here are some well-reviewed cleaners”) means the AI sees your market as a blur. A specific answer (“X is known for move-out cleans, Y focuses on recurring service”) means someone’s content taught it the differences.

If competitors get specific descriptions and you get a generic line, or nothing, the fix is differentiation content: a service page that says exactly what you’re best at, for whom, and how your process differs.

Prompt 3: The question behind the question.

“What should I look for when hiring an [electrician]?”

Nobody asks this prompt expecting a business name, and that’s the point. The AI answers with criteria: licensed and insured, clear pricing, reviews that mention the actual work, examples of past jobs.

Now compare that criteria list to your own website. Every criterion the AI lists that your site doesn’t address is a page you haven’t written yet. These “what to look for” answers are also the pages most likely to get cited, write your own version on your site, in your words, with your examples.

Prompt 4: The local proof test.

“Is there a [detailer] in [your town] with good reviews for [ceramic coating]?”

This one tests depth. It’s not asking about your business in general, it’s asking about a specific service and specific evidence. AI systems can only answer it if someone published that specific combination: your business plus that service plus customer proof.

If the AI names you for your general trade but goes blank on your specialty services, your service pages are too thin. Each service you want to be known for needs its own clear description and its own evidence, photos, reviews that mention it, a page that says you do it.

Prompt 5: The follow-up.

After any of the above, ask: “Tell me more about [your business name]. What do customers say about them?”

This is the entity test: does the AI know your business as a distinct thing, or are you just words on a page? A good answer summarizes your services, your reputation, and your details accurately. A vague or wrong answer, wrong hours, wrong services, a confused description, means your public facts are thin or inconsistent.

Wrong facts are the easiest win on this list. They usually trace back to an outdated directory listing or a half-finished profile. Fix the source, and the answers correct themselves over time.

The fastest version of that fix is the Google Business Profile checklist I wrote — the CLEAR Profile Check, categories, location, evidence, answers, and record consistency, done in one afternoon: https://eric-yim.com/notebook/google-business-profile-setup-for-ai-search. And if you are weighing what AI search rewards against classic SEO, the honest breakdown is here: https://eric-yim.com/notebook/geo-vs-seo-for-service-businesses.

What to do with your results.

Score yourself simply: for each prompt, in each tool, mark named, mentioned vaguely, or absent. That’s your baseline.

Then fix in this order. First, wrong facts: inconsistent name, address, phone, hours, or services anywhere on the web. This is a weekend project. Second, missing evidence: if the AI can’t describe what you do, publish service pages that say it plainly. If it can’t find proof, start the job-to-evidence habit, one photo and one note per job. Third, differentiation: once the basics are covered, publish the content that makes you distinct, the questions you answer better than anyone, the process only you use, the work you’re proudest of.

Re-run the five prompts monthly. You’re looking for movement, not miracles: vague mentions becoming specific ones, wrong facts correcting themselves, your name appearing in a second tool.

I put the full strategy behind this test in my complete guide to AI visibility for service businesses: https://eric-yim.com/notebook/ai-visibility-for-service-businesses.

One warning.

Don’t try to game the test. I’ve seen people ask AI tools about their own business fifty times hoping to “train” it. That doesn’t work, these systems don’t learn your business from your chat history, and repeated self-searching just pollutes your baseline.

The test is a mirror, not a lever. It shows you what the web currently says about your business. Change what the web says, and the mirror changes too.