Does ChatGPT recommend your business? Only if the pages it reads point to you: for "best X near me" questions, ChatGPT, Perplexity and Google's AI answers often search the web first, then name the businesses that lists, reviews, directories and websites describe clearly. To check, ask your buyers' real questions in fresh, logged-out chats, several times each, and count how often you're named. Nobody can guarantee a mention, and the answer changes from run to run.
It usually starts with a customer saying it out loud: “I asked ChatGPT for a good accountant near me and it gave me three names.” Sometimes you were one of them. More often the owner hears this, opens ChatGPT that evening, types “best accountant in Columbus”, and sees two competitors and a firm that closed last year.
Then they ask again and get a different list. Which answer is real? This guide explains how assistants pick the businesses they name, how to test fairly, and what to fix first.
How do AI assistants decide which businesses to name?
An assistant can answer in two ways. It can answer from memory, using what it learned in training, which stops at a cutoff date. Or it can search the web live, read what it finds and write an answer from those pages. A question like “best family dentist in Austin” depends on current, local facts, which is exactly the kind of question that triggers a search.
Here’s what each one says about its own search:
- ChatGPT may search automatically when a question would benefit from current information. It rewrites your question into one or more searches, sends them to search providers it works with, and may use an approximate location from your IP address for local results. Answers that use search can include citations (OpenAI Help Center).
- Perplexity is built around live search. When someone asks a question, it may fetch pages to help answer and link them in the response, and its crawler, PerplexityBot, is what surfaces and links sites in its results (Perplexity docs).
- Google AI Overviews and AI Mode may run several related searches across subtopics (Google calls it “query fan-out”) and show links to the pages they drew on. To be one of those links, a page must be indexed and eligible to show in Google Search with a snippet. Google says there are no extra requirements or special optimizations for these features (Google Search Central).
So the assistant usually doesn’t hold an opinion of your business. It reads a handful of pages and summarizes them. That’s good news, because pages can change. Training data is the slow part: if an assistant answers without searching, it repeats what was written about your market before its cutoff, and only a long track record in respected places moves that.
Why third-party pages decide most answers
Open the sources under any local recommendation and look at what the assistant actually read. You’ll usually see “best of” lists from local magazines and lead sites, directories, review platforms, forum threads, and a few business websites. If you aren’t on those pages, you’re unlikely to be in the answer, even when your own site ranks well.
This is the biggest difference from classic SEO. Your website tells the assistant what you do. Other pages tell it whether you’re worth recommending, and they’re often the pages it quotes. We cover the wider theory of how AI picks sources in our guide to answer engine optimization, and how search engines connect facts about a brand in entity SEO. Here we’ll stay practical: find out where you stand, then fix things in order.
How to check if ChatGPT recommends your business
Most owners test once, from their own account. That tells you almost nothing. A fair test looks like this:
- Write 5 to 10 real buyer questions. Take them from sales calls and contact forms: “Who’s a good estate planning lawyer in Plano for a blended family?” Leave your business name out. Mix broad and specific questions.
- Start clean every time. Use a logged-out window or a temporary chat, with memory off. With memory on, ChatGPT may use what it remembers about you when it writes its searches (OpenAI), so a chat history about your business can make you look more visible than you are.
- Get the location right. Because ChatGPT can use your IP location, testing from your office shows your own neighborhood. If your customers are across town or in another city, put the place in the question.
- Ask each question several times. Run it at least five times per assistant, on different days. One answer is a single draw, not your position.
- Test each assistant separately. ChatGPT, Perplexity, Gemini, Google AI Mode and Claude read different sources and give different lists.
- Log what came back. For each answer, note whether you were named, who else was, whether your site was cited, and which sources were linked. The sources are the most useful column.
In SparkToro's study of 2,961 runs of 12 prompts, ChatGPT and Google's AI had less than a 1 in 100 chance of giving the same list of brands twice, and the same list in the same order less than once in 1,000 runs (SparkToro, January 2026).
The same study found that how often a brand appears holds up much better than where it appears in the list. So count mentions across runs, and don’t read much into “we were third today”. The authors suggest 60 to 100 runs per prompt for a solid number. Five runs won’t give you that precision, but they will show whether you appear often, sometimes or never.
Don't test with "Is [your business] any good?" or "What do you know about [your business]?" Those questions name you, so you'll always be in the answer. They're useful for checking what the assistant says about you, and useless for checking whether it recommends you.
Doing this by hand takes an afternoon. Our AI Visibility Checker does it for you. Its full check (it runs on the free monthly credits that come with an account) asks ChatGPT, Perplexity, Gemini and Claude your questions with web search on, then shows whether each one mentions you, who is named next to you, and which sites were cited. Its “why others are named instead of you” check opens the cited pages from answers that skipped you and tells you, page by page, who is on it and what to do. It’s still a snapshot, so run it again over time.
How to get recommended by ChatGPT: what to fix, in order
Work from the bottom up. There’s no point chasing list articles if the assistants can’t read your site.
Step 1: Let the search crawlers in
Each AI company runs more than one bot, and they do different jobs. OpenAI’s OAI-SearchBot is the one for ChatGPT search: sites that block it won’t appear in ChatGPT’s search answers, though they may still show as a plain link. GPTBot collects training data and has nothing to do with search, and ChatGPT-User visits pages when a user asks, so robots.txt may not apply to it (OpenAI). Perplexity works the same way: PerplexityBot for search results, and Perplexity-User for fetches a user triggers, which generally ignore robots.txt. Google’s AI features follow normal Googlebot rules.
That means you can block training and still allow search:
User-agent: OAI-SearchBot
Allow: /
User-agent: PerplexityBot
Allow: /
User-agent: GPTBot
Disallow: /Both OpenAI and Perplexity say robots.txt changes take up to about a day to apply. Also check your firewall or CDN: bot protection often blocks these crawlers even when robots.txt allows them. Our robots.txt validator shows which crawlers and AI bots can read your site.
Step 2: Make your own pages say what you do, for whom, and where
When an assistant does read your site, it needs plain sentences it can repeat. A line like “We repair and replace central air systems and heat pumps for homes in [your city] and [nearby towns]” can be quoted. “Comfort solutions you can trust” can’t. A common gap is that simple: the homepage never puts the service and the city in the same sentence.
- One page per main service, with who it’s for and the areas you cover.
- The facts buyers compare on: hours, emergency availability, licenses, insurance, languages, what’s included.
- Answers to the questions from your test list, in the words customers use.
- Structured data that matches the visible text. Our guide to schema for AI visibility covers which types help.
Step 3: Get onto the pages that get cited
Your test log now lists the sources the assistants read. Sort them by type, because each needs a different move:
| Cited page | What to do |
|---|---|
| Directory or review site | Claim and complete your listing, then collect reviews there |
| “Best of” list article | Send the author specific facts that earn a place: years, specialties, awards, a reason you suit a type of customer |
| Forum or Reddit thread | Answer helpfully under your real name and say who you are |
| A competitor’s own page | Publish a clearer, more useful page on the same question |
Don't post fake reviews or planted forum comments, and be wary of "top 10" sites that rank whoever pays. Fake reviews break the rules of the platforms the assistants read, and a flagged or removed listing takes you out of the sources entirely.
Step 4: Keep your business details consistent
Your name, address, phone, hours and service list should match on your website, your Google Business Profile and every directory you’re on. When sources disagree, an assistant has to guess, and it may pick the old address or skip you. For local businesses this overlaps with classic local SEO work.
Step 5: Build reviews where the assistants look
Reviews on the sites that keep showing up in your source list count twice: they strengthen that page, and they give the assistant something specific to say about you. Ask every happy customer, reply to reviews, and favor the platforms your assistants cite.
What to expect after you fix things
Be wary of anyone who promises results. Nobody controls these answers, and they vary between runs even when nothing changes. Answers built on live search can shift soon after the cited pages change, so getting onto a source page is often the fastest win. Answers from training data move slowly. Ask the same questions every month and track the share of runs where you’re named.
What to do this week
If you need the ongoing work after that, building mentions and content month by month, see our AI search visibility service.
FAQs
Why does ChatGPT recommend my competitor and not me?
Usually because the pages it read for that question mention your competitor and not you. Look at the sources under the answer: if they’re directories, list articles or review sites, check whether you’re listed and how you’re described. Fixing your presence on those pages is often faster than changing your own site.
Can I pay to be recommended by ChatGPT?
No. ChatGPT now shows ads on its Free and Go plans in a growing list of countries, but OpenAI says ads are labeled as sponsored, shown below the response, and don’t influence ChatGPT’s answers (OpenAI Help Center). An ad sits next to the answer. It doesn’t put you in it.
Does blocking GPTBot stop ChatGPT from recommending my business?
No. GPTBot collects training data. ChatGPT search uses a separate crawler, OAI-SearchBot, and OpenAI treats the two settings independently. If you block OAI-SearchBot, though, your pages won’t appear in ChatGPT’s search answers.
How long does it take to get recommended by ChatGPT?
There’s no fixed timeline. Crawler changes take about a day to register. Answers that come from live search can change soon after the cited pages change, while answers from training data move over months. Track the same questions monthly to see real movement.
Do I need llms.txt or special markup to show up in AI answers?
For Google, no: Google says you don’t need new machine-readable files, AI text files or special markup to appear in AI Overviews or AI Mode. Other assistants haven’t published rules that require it either. Clear pages, open crawlers and a place on cited pages matter far more.
Sources
- Searching the web with ChatGPT, OpenAI Help Center
- Overview of OpenAI crawlers, OpenAI
- Ads in ChatGPT, OpenAI Help Center
- Perplexity crawlers, Perplexity documentation
- AI features and your website, Google Search Central
- AIs are highly inconsistent when recommending brands or products, SparkToro, January 2026
