Now livePublishes straight intoWordPress,Shopify,Wix, Webflow and Ghost. Any other site, hosted on your own domain →

How to Check if ChatGPT Recommends Your Business: A Free Manual Test

31 October 2026. Written by Helio.

To check if ChatGPT recommends your business, write five questions a real buyer would ask, put each one to ChatGPT, Perplexity and Google AI Mode, and record whether you are named and who is named instead. Repeat the same questions monthly in fresh sessions. The test is free, takes under an hour, and a single run means little: the pattern across several months is what tells you where you stand.

Why a manual test beats asking "do you know my business?"

The tempting shortcut is to type your company name into ChatGPT and see what it says. That tells you whether the assistant has heard of you, which is useful, but it is not what buyers do. A buyer types "who fixes boilers in Leeds" or "best florist for wedding flowers near Harrogate". They never type your name, because they do not know it yet.

So the test has to copy the buyer, not the owner. If you want to know whether the facts it holds about you are accurate, that is a separate job, covered in our guide to correcting what ChatGPT says about your business.

Step 1: Write five buyer questions

Five is enough to see a pattern and small enough to repeat every month without dreading it. Aim for a mix:

  • Two "who should I use" questions with your town or area: "Which accountant in Bristol is good for sole traders?"
  • One problem question a buyer has before they know what to buy: "My radiators are cold at the top, who can fix that in Croydon?"
  • One comparison question: "Is it better to use a local IT support firm or a national one for a ten-person office?"
  • One specialist question about the thing you do better than most: "Who does same-day sympathy flowers in York?"

Write them the way a customer would say them out loud, not in marketing language. Ask a couple of recent customers what they searched before finding you if you are unsure. Then freeze the wording. Changing a question between months makes the results impossible to compare.

Rule of thumb: if a question would only ever be typed by you, it does not belong in the test. Every question should be one a stranger with money to spend might ask.

Step 2: Run each question in three assistants

Each tool finds its answers in a slightly different way, so a business can appear in one and be absent from another.

  1. Open ChatGPT in a logged-out window or a temporary chat, so earlier conversations and saved memory do not steer the answer. Make sure web search is switched on.
  2. Open Perplexity and ask the same question. Perplexity shows its sources, which is handy later.
  3. Open Google AI Mode from the Google search page and ask again. If an AI Overview appears on a normal search for the same words, note that too.

Use a private browser window for all three. Ask each question once, cold, and copy the answer before asking a follow-up. Follow-ups such as "any others?" change what you see and are not part of the core test.

Step 3: Record who is named

This is where most people stop at "we were not mentioned" and lose the useful part. The businesses named instead of you are the real finding, because they show what the assistants consider a good answer.

A tracking sheet you can set up today

Open a spreadsheet and create these columns:

ColumnWhat to enter
DateThe day you ran the test
QuestionThe exact wording, copied from your frozen list
AssistantChatGPT, Perplexity or Google AI Mode
Named?Yes, No, or Partly (mentioned but not recommended)
PositionFirst, second, third or later in the answer, if named
Named insteadEvery other business mentioned
Sources citedWebsites the assistant linked or referred to
NotesAnything wrong, odd or worth following up

Five questions across three assistants gives fifteen rows a month. Add a simple tally at the bottom: how many of the fifteen named you. That one number, tracked over time, is your score.

Step 4: Repeat monthly, and expect variation

Run the same fifteen checks once a month, ideally on a similar date. Do not panic if a result flips from Yes to No between runs. AI answers vary for ordinary reasons:

  • The answers are generated fresh each time, so the same question can produce a slightly different list even minutes apart.
  • Live search results change, so the pages the assistant reads this week may not be the ones it read last month.
  • Location affects answers, especially for local questions, so test from roughly where your customers are.
  • The models themselves get updated, sometimes changing their habits overnight.

Because of this, treat one month as weather and three months as climate. If you want a firmer reading, run each question twice in the same session window and count a Yes only if you appear both times.

A single result is a coin toss. A trend across several months is evidence.

Reading your results

Once you have two or three months of rows, look at the "Named instead" and "Sources cited" columns together. Patterns usually show up quickly. The same competitor appearing everywhere often has more reviews, more mentions on local directories or press sites, or pages that answer the exact question asked. If you rank well on Google but still never appear, our article on ranking on Google but not being mentioned by ChatGPT explains the usual causes.

Each question where you are missing is effectively a brief: a page on your site that answers it clearly, plus a reason for other sites to mention you, is the most direct response.

Common questions about checking AI recommendations

How long does a manual ChatGPT visibility test take?

A manual ChatGPT visibility test with five questions across three assistants usually takes thirty to sixty minutes, most of it spent copying business names into the sheet. The first run takes longest because you are writing questions and building the spreadsheet. Later monthly runs are quicker since the questions and columns are already fixed.

Should I use the paid version of ChatGPT for the test?

The free version of ChatGPT is fine for testing, as long as web search is on, because most of your buyers use the free version too. What matters more than the plan is consistency: use the same version, the same logged-out or temporary mode and the same wording each month, so changes reflect the answers rather than your setup.

What if ChatGPT names my business but gets details wrong?

If ChatGPT names your business with wrong details, such as an old address or a service you no longer offer, mark it as Partly in your tracking sheet and note the error. Then check your website, Google Business Profile and directory listings for the outdated information, because assistants often repeat whatever older pages still say.

Where Helio SEO fits in

Helio SEO runs this kind of test automatically, putting buyer questions to an AI assistant with live search, recording who is named, and turning each miss into the next article. If you would like a quick starting point, the free AI visibility check gives you one. If you are happy keeping the spreadsheet above going yourself, you do not need us for this one.

Where Helio SEO fits in. This article was planned, written and published by Helio SEO, doing for our website exactly what it does for customers: a high-quality article every day, targeting a real search your buyers make, in your voice. 7 days free, no card, from £29 a month, cancel anytime.

All articles

A mechanic checking a tablet in his garage

Your first article can be live today

Point Helio SEO at your website and approve your first draft this afternoon. 7 days free, no card. From £29 a month.

Keeping your LinkedIn going too? Helio Posts writes your company page’s posts and publishes them on the days you choose.