Check Whether ChatGPT Recommends Your Business (and Fix It If It Doesn't)

A customer asks ChatGPT for "the best bookkeeping service for small agencies" and gets five names back. Is yours one of them? Here's how to find out in about 30 minutes, without any SEO background, and what to do with the answer.

Why this is worth half an hour

A lot of buying research now starts in a chat window instead of a search box. Somebody types a question, reads a short answer, and walks away with a shortlist of three to five brands. If you're not on it, you never hear about the lost lead. There's no "page two" to scroll to.

The tricky part is that you can't see this happening. Google Search Console shows you clicks and rankings. Nothing equivalent tells you that Perplexity recommended your competitor 40 times last week. So you have to go look.

Step 1: Write down the questions your buyers actually ask

Don't start with keywords. Start with sentences. Open a blank doc and write 10 to 15 questions a real customer might type into ChatGPT before they ever contact you.

A mix works best:

• Category questions. "What's the best project management tool for a 10-person design studio?"
• Problem questions. "How do I stop clients from paying invoices late?"
• Comparison questions. "Is [your competitor] worth it, or are there cheaper alternatives?"
• Location or niche questions, if they apply. "Good HVAC repair companies in Tampa."

Your sales or support inbox is a goldmine here. Copy the wording customers use in their first email, because that's usually how they phrase it to an AI too. If you can't think of 10, ask whoever answers the phones. They'll have 30.

Step 2: Run the questions through the engines, logged out

Use a private or incognito window and, where you can, stay logged out. ChatGPT with memory turned on knows you run your business and may mention you just because you've talked about it before. That's flattering and useless.

Run each question in at least these four:

1. ChatGPT
2. Perplexity
3. Google (look at the AI Overview at the top of results, when one shows up)
4. Gemini

Claude and Copilot are worth adding if your buyers are likely to use them, for example if you sell to developers or to companies on Microsoft 365.

One run isn't enough. These answers change between sessions, sometimes a lot. Ask each question two or three times, a few hours apart if you can. A brand that shows up three times out of three is in a very different position from one that showed up once.

Step 3: Log what came back in a spreadsheet

Make a simple sheet with these columns:

• Question
• Engine
• Run #
• Were we mentioned?
• Who was mentioned
• Sources cited

The last column is the one people skip, and it's the most useful. Perplexity and Google's AI Overviews show their sources openly. ChatGPT often does too when it browses. Write down every domain that gets cited. After 40 or 50 rows a pattern usually appears: the same handful of review sites, industry blogs, comparison articles and directories keep showing up as sources.

Those sites are your map. They're where the engines go to learn who's good in your category.

Step 4: Read the answers, not just the yes/no

Count your mentions, sure. But also read what's actually said:

• Are you described correctly? Old pricing, a product you discontinued and the wrong city are all common. An AI repeating a 2023 price is a real sales problem.
• What do you get compared to? If ChatGPT keeps placing you next to enterprise tools when you're a budget option, your positioning isn't coming through anywhere it reads.
• Who wins the questions you lose? Look at the brands that show up when you don't. Then check their sources. Often they're not better companies, just better documented ones.

Also ask the engines about you directly: "What do people say about [your brand]?", "[your brand] pricing", "[your brand] vs [top competitor]". These are the questions someone asks right before they buy. If the answer is thin or wrong, that's the first thing to fix.

Step 5: Fix the stuff you control first

Some of what you find can be fixed on your own website this week.

Make your pricing and basics easy to find. A plain pricing page, a clear "who it's for" line on the homepage, and your location and service area written out as text, not hidden in an image. Engines can't quote what they can't read.

Answer the questions from Step 1 on your site. Not as a stuffed FAQ. Write a proper page for the three or four questions where you lost. If people ask "is X worth it for small teams", write the honest answer with numbers in it.

Write a comparison page. "[You] vs [Competitor]" pages get cited a lot, because they answer exactly what buyers type. Be fair about where the other option is better. A page that only praises you reads like an ad, and nobody quotes ads.

Clean up your listings. Google Business Profile, G2, Capterra, Trustpilot, Yelp, industry directories, whatever showed up in your sources column. Same name, same description, current prices.

Step 6: Get mentioned on the sites the engines already trust

This is the step most small businesses skip, and it's usually the one that moves things.

Remember the sources column? Engines lean heavily on what other sites say about you, not just what you say about yourself. An Ahrefs study of 75,000 brands, published in 2025, found that brand mentions across the web correlated more strongly with visibility in Google's AI Overviews (0.664) than backlinks did (0.218). That's a correlation, not proof. Big brands get more of everything. But it lines up with what you'll probably see in your own spreadsheet: the brands that win are the ones talked about on third-party sites.

So the job becomes getting your name onto the kinds of pages that keep turning up as sources. In practice that means:

• Pitching yourself for "best X for Y" roundups in your niche
• Contributing a useful guest article to an industry blog
• Asking happy customers to leave reviews on the platforms that got cited
• Getting listed in the directories and comparison sites from your list

It's slow and a bit tedious. Pitches get ignored. Editors want something different from what you sent. That's normal.

Step 7: Re-check every month

Run the same question list again a month later, same engines, same method. Compare. You're not looking for a perfect score. You're watching whether your mention rate on the important questions goes up and whether the descriptions get more accurate.

Keep the question list fixed for at least three months, or you won't be able to tell what changed.

When to stop doing this by hand

The manual version works fine for a first look. Fifteen questions, four engines, three runs each is 180 answers. That's an afternoon. Doing it every month, for every market you sell in, plus chasing placements, is where it falls apart for most small teams.

If you'd rather hand the tracking and the follow-up off, you can get cited by AI engines via RankCited.com. It runs your buyer questions through ChatGPT, Perplexity, Gemini and Google's AI answers, shows which competitors appear when you don't, and turns the gaps into an action plan: pages to write or fix, plus paid placements on publications in your niche that its team arranges for you. There's a free monthly report with no card required, so you can compare its results to your own spreadsheet before deciding anything. Paid placements are disclosed, sponsored articles, so treat them as one part of the mix and keep doing the earned stuff (reviews, roundups, honest comparison pages) alongside them.

Whichever way you go, the first move is the same. Open an incognito window, ask ChatGPT the question your best customer asked you last month, and see who it recommends.