Can ChatGPT recommend my business? A 5-minute test

Yes, ChatGPT can recommend your business, and it already recommends someone's. Ask it for "a good accountant near me" or "the best CRM for a small trades business" and it names names. Whether it names yours depends on what it can find, read, and verify about you online. For most small businesses the honest answer today is: it doesn't know you exist.

That's the part worth fixing, and it's more fixable than most owners assume.

How does ChatGPT decide which businesses to name?

When you ask ChatGPT a question that needs current information, it runs a live web search in the background (powered by Bing, through OpenAI's Microsoft partnership), reads a couple of dozen results, and then re-ranks them by its own criteria before writing an answer. It leans towards sources it can verify: consistent business details, real reviews, pages that state plainly what the business does and for whom.

Notice what's missing from that list: your Google ranking. The AI tools draw on search indexes, but they don't simply repeat page one. BrightEdge's tracking of Google's own AI Overviews found that only around 17% of the sources cited in AI answers also rank in the organic top ten for the same query. Being invisible in AI answers and being invisible in Google are related problems, but they are not the same problem.

There's proper research on what does move the needle. A Princeton-led study on generative engine optimisation tested 10,000 queries against the AI engines. It found pages could lift their visibility in AI answers by up to 40% with unglamorous changes: adding sourced statistics, citing references, and including quotable, clearly attributed statements. The same study found the biggest gains went to pages that weren't ranking first anyway. If you're a small firm that was never going to outrank the big directories, that's unusually good news. The AI engines reward being checkable more than being big.

How do I test my own visibility?

Don't take my word for any of this. Run the test yourself; it takes five minutes.

1. Open ChatGPT (and Claude or Gemini if you have them) and ask the question a customer would ask: "Who would you recommend for [what you do] in [your area]?"
2. Ask the follow-up: "What do you know about [your business name]?" and see whether what comes back is accurate, outdated, or blank.
3. Ask it about your biggest competitor, and note the difference.

Three results are possible. You're recommended: rarer than you'd hope. You're known but not recommended: the AI has read about you but found nothing that makes you the confident answer. Or you're invisible: it has nothing, or worse, it guesses. Each of those points at different fixes, which is exactly why the diagnosis is worth doing before spending anything.

I'm not preaching from a pedestal here. When we ran our own audit tooling over this site in July, it flagged every single page as under-structured for AI parsing, and Google had indexed just four of our twenty-eight pages. We've been fixing both since. The test is humbling for almost everyone who takes it.

Should a small business care whether ChatGPT recommends it?

The audience is no longer a rounding error. OpenAI reported 900 million weekly active users in February 2026. Meanwhile the old route to being found is quietly narrowing: SparkToro's analysis of Similarweb data found 68% of Google searches now end without a click on any website, up from 60% in 2024. More questions are being answered on the results page, or in a chat window, without anyone visiting anyone.

And the visitors who do arrive from AI tools behave like better prospects. Adobe's analytics across US retail sites measured AI-referred traffic up 138% year on year in May 2026, and found those visitors convert meaningfully better than average. That fits common sense: someone who arrives after an AI walked them through their options has already done their shortlisting.

Being an accountant, I'll also give you the counterweight. A 2026 consumer survey by Fractl found trust in AI search results has fallen sharply even as usage climbs, and Yext's research found that over 90% of people verify an AI recommendation before acting, most commonly by visiting the business's website. So an AI mention doesn't replace your website; it sends sceptical, high-intent people straight to it. If the site they land on is thin or stale, the recommendation dies there.

What makes a business recommendable?

Pulling the research and the mechanics together, the businesses that get named share a few habits, none of which involve tricks:

• Their basic facts agree everywhere. Name, location, services, hours: consistent on the website, Google Business Profile, and the directories. AI engines cross-check, and inconsistency reads as unreliability.
• Their site answers questions in plain sentences. A page that says "we do X, for Y businesses, in Z" gives an AI something it can quote. A page of vague brand copy gives it nothing.
• They show evidence. Numbers, named sources, real case studies. The Princeton study's single most effective tactic was adding sourced statistics.
• Their site lets AI crawlers in. Plenty of websites block the AI bots in robots.txt without anyone ever deciding to, and a blocked crawler means an invisible business.

None of this is exotic. It's the online equivalent of keeping tidy books: boring, verifiable, and it compounds.

What would I actually do first?

Run the five-minute test above before anything else. If you're already recommended, take the afternoon off. If you're not, the sensible next step is a proper diagnosis rather than guesswork, which is precisely what our Search & AI Visibility audit does: it checks how your site performs in Google and across the AI engines, with the evidence behind every finding, so you fix what's actually broken rather than what a blog post (including this one) guessed might be.

Either way, ask the machines about yourself this week. The answer is usually clarifying. Sometimes it's the most useful bad news you'll get all quarter.