How to check if ChatGPT recommends your business

You can check whether ChatGPT recommends your business in about ten minutes, for free. Run three types of prompts (a recommendation prompt, a brand prompt, and a decision prompt) in ChatGPT, Gemini, Perplexity, and Google's AI Mode, then note three things: whether you appear, what the AI says about you, and which sources it cites. That is the whole test.

Most owners have moved past wondering whether AI search matters. The data settled it. The question now is more uncomfortable: are you in the answers or not? Almost nobody has actually looked. This article is the Monday-morning version of looking, plus the fix list for whatever you find.

Why check your AI visibility at all?

Because a real share of your next customers are already asking AI who to hire, and no report you currently get shows what it tells them. BrightLocal's 2026 Local Consumer Review Survey found 45% of consumers used AI tools like ChatGPT, Gemini, and Google's AI Mode to find local businesses in the past year, up from 6% the year before. That makes AI the third most-used discovery channel for local businesses, behind only Google and Facebook.

ChatGPT leads the pack: 31% of consumers asked it for a business recommendation in the past year, with Google's AI Mode at 23% (BrightLocal, 2026). And the people asking are not skimming. Among active AI users, 63% trust the recommendations they get, while 88% still verify against reviews and sources afterward (BrightLocal, 2026). The AI answer has become the front door; your reviews are the second check, not the first.

The traffic is small but it closes. An analysis of 94 ecommerce sites found ChatGPT referral traffic converted 31% higher than non-branded organic search (Search Engine Land, 2026). A person who asked an assistant "who should I call" and clicked through arrived pre-sold. Ten minutes to find out whether that person can ever reach you is cheap.

What is the 3-prompt test?

It is three questions that mirror the three moments a buyer meets your business inside an AI chat: discovery, vetting, and the final call. Run each one in a fresh chat, phrased the way a customer would say it, not the way a marketer would.

• The recommendation prompt. "Who does the best furnace repair in Etobicoke?" or "I need a family dentist near Leslieville, who should I look at?" This is the money prompt. It shows whether you exist in the consideration set at all.
• The brand prompt. "Is [your business name] any good?" This is what the AI says when someone checks you out after a referral or a truck sighting. You are testing accuracy here, not presence.
• The decision prompt. "[Your business] vs [your top competitor], who should I hire for X?" Late-stage, high intent. You want to know which way the AI leans and what evidence it leans on.

Run two or three variations of each prompt, because AI answers shift between runs and that is normal. You are looking for the pattern, not a single verdict. Log it somewhere dumb and simple: a spreadsheet with the platform, the prompt, whether you appeared, what was said, and what got cited. That log becomes your baseline for every future check.

Where should you run the prompts?

Four places minimum: ChatGPT, Google's AI Mode, Gemini, and Perplexity. Add Microsoft Copilot if you have five more minutes. Each one reads different sources, so a clean bill of health in one says nothing about the others.

• ChatGPT is the volume leader (31% of consumers used it for recommendations per BrightLocal, 2026) and pulls heavily from Bing's index, which is why Bing Places suddenly matters again.
• Google AI Mode and AI Overviews sit inside the search results your customers already use, drawing on Google's own local data. 23% of consumers used AI Mode for recommendations this year.
• Gemini reads Google's ecosystem, including your Business Profile, so it doubles as a check on your GBP data.
• Perplexity shows its citations openly on every answer, which makes it the best place to see exactly which pages and platforms are feeding the machines.

If Google's AI surfaces feel like a separate beast, they are. We covered what AI local packs changed and how AI search is reshaping local visibility in earlier pieces. For this test, treat them as one more place to run the same three prompts.

Which free tools automate the check?

Two are worth your time before anyone asks for a credit card. Ahrefs' AI Visibility Checker runs without a signup: it queries ChatGPT, Gemini, Perplexity, Copilot, AI Overviews, and AI Mode using prompts derived from real search behavior, then shows where your brand gets mentioned and which domains get cited alongside you. Semrush's AI Search Visibility Checker adds a visibility score and a competitor comparison, useful for seeing who is eating the mentions you want.

The tools have one blind spot, and it matters for local. They test broad, category-level prompts at scale. They will not ask "emergency plumber near Coxwell station who answers on Sundays" the way your actual customer does. So use the tools for the wide sweep and the manual 3-prompt test for the queries that pay your invoices. Ahrefs' own tracking guide makes the same point: real prompts beat synthetic ones.

What do the results actually mean?

Every result lands in one of three buckets: absent, mentioned but wrong, or mentioned and accurate. Before you read yours, calibrate. SOCi's 2026 Local Visibility Index, which analyzed over 350,000 business locations, found ChatGPT recommends just 1.2% of them, against a 35.9% average appearance rate in Google's local 3-pack. Getting recommended by AI is 3 to 30 times harder than ranking in traditional local search.

So absence is the default, not a death sentence. If the AI never names you, it almost certainly is not naming most of your competitors either, which means the slot is still open. Mentioned but wrong is more urgent than it sounds: an assistant confidently telling buyers your old address or a service you dropped is actively costing you jobs. And mentioned accurately means you are one of the few being handed pre-sold buyers, and your only job is not to lose it.

What do you fix for each outcome?

You fix the sources, because sources are all an AI assistant reads. Yext analyzed 6.8 million AI citations across ChatGPT, Gemini, and Perplexity and found 86% come from places businesses already manage: first-party websites (44%) and listings and directories (42%). Reddit and forums, for all the noise about them, accounted for about 2% once location and intent were applied.

If you are absent, the machines have nothing to work with. Fill your Google Business Profile to the bottom: categories, services, attributes, photos of real jobs. Claim and complete Bing Places, since ChatGPT leans on Bing's index; Whitespark's research found Facebook and Yelp reviews feed that index far more than most owners assume. Keep your name, address, and phone identical everywhere. Then give the assistants something worth quoting: pages that answer real buyer questions directly, marked up with schema, the approach we broke down in the AEO guide.

If you are mentioned but wrong, hunt the stale source. Old hours on a directory you forgot existed, a dead landline on an unclaimed listing, a services page you never updated after dropping a service. Fix the data where it lives and the answers follow, usually within weeks. If you are cited and accurate, keep feeding the machine: steady detailed reviews (here is the system), fresh content, and a monthly re-run of the same prompts so you notice drift before your customers do.

This is the same structural work behind our rebuilds, not a separate "AI SEO" product. When RMCM rebuilt E&M Equipment's site, the SEO health score went from 31 to 90, and the changes that moved it (clean headings, direct answers, schema, consistent business data) are exactly what makes a site quotable to a machine. Build for the buyer's question and every engine benefits.

Frequently asked questions

You cannot manage a channel you have never looked at

You check your Google reviews. You glance at your rankings. Meanwhile the third-biggest discovery channel for local businesses has been answering questions about you all year, and most owners have never once read what it says. Ten minutes fixes that. Not the visibility itself, the not-knowing.

So run the three prompts this week and put a monthly repeat in the calendar. If the answers show gaps, that is the exact terrain of an RMCM local SEO engagement: the profile, the listings, the reviews, and a site the machines can quote. The free audit will show you in 30 seconds how your website reads to the systems doing the recommending.