Why ChatGPT Doesn't Recommend Your Business (and How to Check)

You've probably done it already: opened ChatGPT, typed "best [what you do] in [your town]", and looked for your name. Maybe you found it. More likely you saw two competitors and a directory site, and wondered what they did that you didn't.

This article covers two things. First, how to check properly, because one quick question tells you less than you'd think. Second, the most common reasons a solid business gets left out, and what to do about each one.

First, check properly

AI answers aren't a fixed ranking. Ask the same question twice and you can get different names. That's not a rare glitch; it's how these systems work. In one published test across 200+ US markets, researchers asked ChatGPT the same "who is the best [service] in [city]" question twice, 4 to 9 days apart. Only 48.2% of the businesses recommended the first time were recommended again.

So one screenshot, good or bad, doesn't mean much. Here's a test that does.

Step 1: Write down the questions your customers actually ask

Not keywords. Questions, in the words a customer would use. Look at your recent calls, emails and contact form messages. You'll usually find five to fifteen distinct ones. Mix three types:

• Best-of: "Who's the best family dentist in Fountain, CO?"
• Problem: "Who can fix a furnace that keeps shutting off in Monument?"
• Specific fit: "Which accountants near Colorado Springs work with small construction companies?"

Two rules: don't put any business name in these questions, yours or a competitor's, and always include the place, the way a real customer would.

Step 2: Ask them in a clean session

Log out, or use a private browser window. A logged-in assistant may remember things about you and skew the answer. Ask each question in ChatGPT, and if you have time, in Perplexity, Gemini, Claude and Google (look for the AI Overview at the top, or try AI Mode). Ask each one at least three times, ideally on different days.

Step 3: Log what you see

A simple spreadsheet works. For each answer, write down:

• Were you named? In what position?
• Who was named instead?
• Which sources were cited or linked? (Perplexity and Google show these clearly; ChatGPT shows them when it searches the web.)

Count how many answers named you out of the total. That's your appearance rate, and it's the number worth tracking over time.

Step 4: Ask about yourself by name

Now ask directly: "What do you know about [your business] in [your town]?" and "What are [your business]'s hours and phone number?" Check every fact. Wrong hours, an old phone number, or services you stopped offering are all clues about where the engine is getting its information.

The usual reasons you're left out

Once you've run the test, the reasons tend to fall into a few groups. Most businesses have more than one.

1. The engine can't read your site

AI assistants that search the web send crawlers to fetch pages. OpenAI documents its crawlers on its developer site (OAI-SearchBot is the one used for ChatGPT search results), and Perplexity documents PerplexityBot. If your robots.txt, security plugin or firewall blocks them, your own site can't be used as a source. This happens more than you'd expect, often because a hosting provider turned on "block AI bots" by default.

Check: open yoursite.com/robots.txt and look for Disallow rules under those bot names. Ask your host whether any bot-blocking feature is switched on.

2. Your site doesn't clearly say what you do and where

A lot of small business sites have one "Services" page with a list of bullet points and a stock photo. An engine trying to answer "who does emergency drain cleaning in Security-Widefield?" needs a page that plainly says you do emergency drain cleaning, in Security-Widefield, and gives some detail. If that sentence doesn't exist anywhere, you're asking the engine to guess.

Fix: one page per main service. Open with two sentences that answer the obvious question: what it is, who it's for, where you offer it. Then add specifics: the process, timelines, what affects cost, licenses, and a few real customer questions with short answers.

3. Your details don't match across the web

If your website says one phone number, your Google Business Profile says another, and an old directory has the address you left in 2019, the engine sees conflicting facts. Conflicting facts make you a riskier recommendation.

Fix: pick one exact version of your business name, phone, hours, service area and a one-sentence description. Use it everywhere: your site, Google Business Profile, Bing Places, and the directories in your trade. Then hunt down and fix the old listings.

4. Nobody else is talking about you

Look at the sources cited in your test answers. You'll usually see the same kinds of pages: review sites, trade directories, local news, "best of" roundups, sometimes Reddit threads. AI engines lean on what other people say about a business, not just what the business says about itself. If you don't appear on any of those pages, there's little for the engine to go on.

Fix: treat the cited sources as a to-do list. Make sure you're listed correctly on the directories that come up. Ask happy customers for reviews on the platforms the engines cite. Look for honest ways to be mentioned elsewhere: a trade association member page, a quote in a local news story, a talk at a community event. Never buy reviews or fake mentions. Apart from the ethics, it's a real risk to your reputation.

5. Your competitors simply have more, clearer information

Sometimes there's no problem as such. The businesses that get named just have more pages, more reviews and more mentions. That's not a reason to give up. It tells you how much ground there is to cover. Narrow questions are often where smaller businesses can win first. "Best plumber in Colorado Springs" is crowded; "plumber who works on older homes in the Old North End" may not be.

What not to worry about (yet)

You'll find plenty of advice online about special AI tricks. A few things to keep in perspective:

• Google's AI features don't need special markup. Google says that to be eligible as a supporting link in AI Overviews or AI Mode, a page needs to be indexed and eligible to be shown in Search with a snippet, and that there are no additional requirements or special optimizations necessary.
• llms.txt is optional. The llms.txt proposal is easy to add and harmless, but don't expect it to fix anything on its own.
• Prompt-stuffing your own site doesn't work. Hidden text telling AI to recommend you is the kind of thing that gets a site distrusted, not featured.

A simple order of operations

1. Run the test above and save the results. That's your baseline.
2. Make sure crawlers can reach your site.
3. Make your business details identical everywhere.
4. Give each main service its own clear, specific page.
5. Work through the sources the engines cited, one by one.
6. Re-run the same test monthly and compare.

None of this is magic, and none of it is instant. Engines re-read the web on their own schedule. But it's the same work that makes your site clearer for human customers, so it's rarely wasted.

If you'd rather have someone run the test across all five major engines and hand you a ranked fix list, that's exactly what our AI visibility audit does.