TL;DR
- ChatGPT tends to name the business it can most easily find, confirm and match to the question. That is not always the best business.
- OpenAI documents that ChatGPT can rewrite a 'near me' request into a search that includes the user's general location, and share that location with search providers such as Bing.
- Check six gaps: crawler access, matching facts, a page that answers the question, third-party sources, location, and normal answer variance.
- Skip llms.txt files, 'AI schema' and pay-to-submit offers. Google says no special AI files or markup are needed for its AI features.
- Test the same prompt over time, log the sources, and treat any promise of a guaranteed AI mention as a red flag.
Why does ChatGPT recommend my competitor and not me?
Usually because your competitor is easier for ChatGPT to find, confirm and match to the exact question. ChatGPT can rewrite the request into web searches, use the searcher's location, and build its answer from the pages and listings it retrieves. If your site is blocked, your facts conflict, or no page matches the question, a rival fills the gap.
That does not mean they are better at the work. As one guide on this topic puts it, the assistant is not judging which company is better. It is judging which company it can describe confidently, and those are different things.
This matters more each year. In BrightLocal's 2026 consumer survey (a vendor survey), 45% of US consumers said they used ChatGPT or other generative AI tools for local business recommendations in the past 12 months, up from 6% the year before .
How does ChatGPT decide which business to name?
Short answer: it often searches the web first, then writes from what it finds. OpenAI has not published a ranking formula, but its help pages explain the steps.
It searches, then writes
Per OpenAI's search help page, ChatGPT may decide that current web information would improve a response. It can then turn the request into one or more search queries, retrieve results, and use them to write an answer with links to sources. After reviewing the first results, it may send more specific follow-up queries to other search providers.
"Near me" becomes "[service] in [city]"
This is the part most owners never hear about. OpenAI's web search help page says ChatGPT collects general location information based on your IP address and may share that general location with third-party search providers . OpenAI's own example: a user types "What are some good restaurants near me?", ChatGPT figures out they are in the San Francisco area, and it may rewrite the prompt into the search query "top restaurants San Francisco."
Think about what that means for you. Your customer typed "near me." The search ChatGPT actually ran may look like "top plumbers [city]." The pages that win that kind of search are often "best of" lists and directories, not your homepage. That alone can explain why a rival gets named.
Users can also share exact location. According to OpenAI's release notes, as of March 26, 2026, users can choose to share their device location for things like local recommendations, and sharing is optional and off until they turn it on . On iOS and Android, results may also include a map.
Bing is one of the places it looks
OpenAI's enterprise help page is more specific than the consumer page. It says ChatGPT may share approximate user location derived from an IP address with Bing and specialized data providers to improve the accuracy of results . So if your business is weak or wrong on Bing, that gap may show up in ChatGPT too.
What OpenAI does not tell you
OpenAI has not published the signals it uses to pick one business over another. A trade site quoting OpenAI's help center reports this line: "ChatGPT ranks search results using multiple factors intended to help users find relevant, reliable information. Placement is not guaranteed." OpenAI also warns that search results and citations can be incomplete, outdated, or incorrect. If someone claims to know the exact formula, they are guessing.
The six gaps at a glance
Here is the short version. Each gap gets its own section below.
| Gap | What it looks like | First check |
|---|---|---|
| 1. Blocked crawler | ChatGPT never cites your site | robots.txt and host bot settings |
| 2. Facts disagree | Different name, phone or hours online | Google, Bing, Apple, Yelp side by side |
| 3. No matching page | Vague service pages | One clear page per service |
| 4. Missing from sources | Lists and directories name rivals | Open the cited sources |
| 5. Location | Rival is closer to the searcher | Name the area in your test |
| 6. Normal variance | You appear some days, not others | Repeat and log the prompt |
Gap 1: Can ChatGPT actually reach your site?
Check this first because it is the cheapest gap to fix. OpenAI runs separate bots for separate jobs. Per OpenAI's crawler documentation, OAI-SearchBot is used to surface websites in ChatGPT's search features. Sites that opt out of it will not be shown in ChatGPT search answers, though they can still appear as navigational links.
The training bot is a different setting. OpenAI says each setting is independent: a site can allow OAI-SearchBot to appear in search results while disallowing GPTBot to keep its content out of model training .
| Bot | Job | If you block it |
|---|---|---|
| OAI-SearchBot | ChatGPT search | Not shown in ChatGPT search answers |
| GPTBot | OpenAI training | Separate from search |
| ChatGPT-User | User-requested fetches | OpenAI says robots.txt rules may not apply |
| Claude-SearchBot | Claude search | Anthropic says visibility may drop |
What to do:
- Open yourdomain.com/robots.txt. Look for "Disallow: /" under OAI-SearchBot, or under "User-agent: *" with no exception.
- Ask your host or CDN what its "block AI bots" setting actually blocks. OpenAI recommends allowing OAI-SearchBot in robots.txt and allowing requests from its published IP ranges . A firewall can block the bot even when robots.txt says yes.
- Allow time for the change to be read. OpenAI says a robots.txt change can take about 24 hours to affect search results.
Whether to allow training bots is your call. Allowing the search bots is what keeps you eligible. My full walkthrough is in how to allow AI crawlers in robots.txt.
Gap 2: Do your business facts match everywhere?
If your name, address, phone and hours differ across the web, an assistant trying to confirm you finds conflicting evidence. A competitor with identical details everywhere is easier to confirm. The Answer Engine makes a fair point here: one weak link, such as an outdated bio, a vague service page or conflicting phone numbers, can tip the recommendation to your competitor .
Common drift looks like this:
- "Smith Roofing LLC" on Google, "Smith Roofers" on Bing
- An old suite number on one directory
- A call-tracking number on the website, a different number on Google
- Hours that changed last year and never got updated elsewhere
- A service area that no longer matches the work you take
Line up your Google Business Profile, Bing Places, Apple Business Connect, Yelp and your industry's main directory. Fix them until they say the same thing. Google's AI features guidance also points to keeping Business Profile information current. For Apple, see Apple Business Connect setup.
Gap 3: Does any page on your site answer the question?
If ChatGPT rewrites the request into a search like "emergency AC repair [city]" and no page of yours plainly matches it, a competitor's page gets used. Read your own site like a stranger. Does it say what you do, where, and for whom? Or does it say "quality solutions for all your needs"?
Compare:
- "Same-day water heater replacement in [city] and [nearby town]. Gas and electric. Call for a price."
- "Excellence you can count on."
Only the first one answers a question. Each service page should name the service, the area, who it is for, and what to do next. Put it in real text, not inside images or sliders.
There is some research support for clear, evidence-rich pages. The Princeton-led GEO paper found that adding citations, statistics and clean structure to a page could "boost the visibility of these sources by up to 40%" in AI-generated answers . That was a research test setup, not a published ChatGPT ranking rule, so treat it as direction, not a guarantee.
Your own site still counts for a lot in local answers. In BrightLocal's vendor study, 93% of all unique domains used as sources were business sites, and they made up 42% of all citations.
Gap 4: Is your competitor in the sources ChatGPT reads, and you are not?
Often, yes. ChatGPT names businesses it finds in the pages it pulls. If those pages are lists and directories that include your rival and skip you, you lose before your own site is even read.
Pomo's guide gives a good diagnostic: ask the engine the exact question a buyer asks, then look at which sources it cited. Often the cited page is a listicle on a site you have never heard of, or a competitor's comparison post. Ahrefs, an SEO software vendor, said in a post on X that "best X" listicles were 43.8% of the page types ChatGPT cited in its research. I have not reviewed the underlying data, so treat that as a vendor claim.
For local businesses, directories and review sites show up a lot. BrightLocal's study of about 1.97 million local AI citations (a vendor study) found:
- Google Business Profile made up 28.5% of all citations , counted across Google AI Overviews, AI Mode and ChatGPT together.
- Yelp made up 8.53% of all citations, with increased usage on ChatGPT results.
- Legacy directories like the Better Business Bureau, Angi and MapQuest were regularly cited.
What about reviews? The Answer Engine argues that ChatGPT cannot see reviews the way humans do. I would not lean on that. Review platforms like Yelp show up as cited sources, so your profile there is part of what gets read. Ask real customers for honest reviews and reply to them.
Never fake it. Manufactured reviews and astroturfed community posts are detectable and against most platforms' terms , as Pomo notes.
Gap 5: Is your competitor just closer to the person asking?
Sometimes. Since ChatGPT can fold the searcher's general location into its search, a rival near that person may be favored for a "near me" question. You cannot move your building. You can be clear about the areas you really serve.
Also, a strong map ranking does not carry over cleanly. In BrightLocal's study, Google Maps mentioned a tracked business in 66% of searches vs 32 to 38% for AI surfaces (AI Overview 38%, ChatGPT 33%, AI Mode 32%) . A separate SOCi figure, reported by a marketing agency, found ChatGPT recommended 1.2% of multi-location brand locations, compared with a 35.9% appearance rate in Google's local 3-Pack . Both are vendor numbers. The direction is the point: winning the map does not mean winning ChatGPT. More in ChatGPT vs Google for local search.
Gap 6: Is it a gap at all, or just a different answer today?
It may just be variance. Answers shift from run to run. BrightLocal found that the list of recommended businesses changes when the same local prompt is run repeatedly, but businesses reappear 50%+ of the time. It also found that ChatGPT mentions the most businesses, but the average across platforms is 2 to 4.
Short lists plus shifting answers mean one bad result tells you very little. A log of repeated tests tells you a lot.
What does not work?
Several offers sound technical but have no support in the official docs.
- llms.txt files. Per summaries of Google's guidance, you do not need to create new machine-readable files, AI-only text files, or special schema.org structured data. More in llms.txt for small business.
- "AI schema." Google says a page must be indexed and eligible to be shown in Google Search with a snippet to be a supporting link in AI Overviews or AI Mode, and there are no additional technical requirements. Normal structured data is fine when it matches the page. See local business schema for AI search.
- "Submit to ChatGPT" services. OpenAI's site owner guidance, as quoted by eSEOspace, is to allow OAI-SearchBot and confirm that your host or CDN allows traffic from OpenAI's published searchbot IP addresses . Nothing in the OpenAI docs I checked describes paying to be named in an answer.
- Fake reviews and fake community posts. Covered above. Skip them.
The Google docs cover AI Overviews and AI Mode, not ChatGPT. I include them because the pattern matches: crawl access, clear content and accurate listings, not a magic file.
How do I test what ChatGPT says about my business?
Run the same prompt on a schedule, name the area, and log the sources. One answer is an anecdote. A log is evidence.
- Write the customer's prompt. Use their words, like "Who is the best family dentist in [city]?"
- Name the area. Do not rely on location guessing. OpenAI suggests that if a result looks wrong, you can ask ChatGPT to search again using a specific source, date, or location.
- Open every source. Responses that use web search may include citations, and you can select a citation to open its source. Mark each one: yours, a competitor's, or a third party's.
- Log it. Date, exact prompt, area, businesses named, sources cited.
- Repeat. Same prompt, same wording, on a schedule.
- Vary the phrasing. Try "near me," a named city, and a specific service.
The goal is not to chase one answer. It is to learn which sources ChatGPT leans on in your category, then fix the ones that are wrong or missing you. A step-by-step version is in how to check if AI recommends your business.
What should I fix first?
Work from cheapest to slowest. Nothing here guarantees a recommendation. It makes you easier to find and confirm.
- Check robots.txt and your host's bot settings for OAI-SearchBot and the other search bots.
- Claim or recover your Google Business Profile. Fix the category, hours and service area.
- Set up Bing Places and Apple Business Connect, and make them match Google exactly.
- Fix your Yelp profile and your industry's main directories.
- Rewrite thin service pages so each answers one customer question in plain text.
- Add structured data only where it matches the visible page.
- Ask happy customers for honest reviews, and reply to each one.
- Find the "best of" lists ChatGPT cites for your category, and ask fairly to be considered.
- Run the test above and keep the log.
For the full list, see the AI visibility audit checklist.
Does this apply to Claude, Gemini, Grok, Perplexity and Google AI Mode?
Partly. The groundwork is shared: reachable pages, matching facts, clear descriptions. The details differ by company.
- Google AI Overviews and AI Mode. Google says AI Mode and AI Overviews may use different models and techniques, so the responses and links they show will vary. Trade press also reports Google now says a site "must be included in Search generative AI features in Search Console" to be eligible. Check that setting. See Google AI Mode for local business and getting found in AI Overviews.
- Claude. Search Engine Journal reports that Anthropic's crawler documentation lists separate Claude bots for training, search indexing and user requests, with visibility tradeoffs when blocked. See how to get found in Claude.
- Gemini, Grok and Perplexity. I did not verify their official crawler or ranking docs for this article, so I will not guess. Read each company's own documentation before blocking anything. Start with Gemini, Grok and Perplexity.
What I can and cannot promise
I'm John Alewine, and I run Get Found with AI. We help businesses across the US get found, cited and recommended in ChatGPT, Claude, Gemini, Grok and Perplexity, plus Google AI Overviews and AI Mode. You can see how it works and our services.
Here is what I can't do: I can't promise any placement. No one can guarantee a ChatGPT mention, an AI citation or a timeline, and OpenAI itself says placement is not guaranteed. I also have no client results to publish here, and I would rather say that than invent a number. If you are comparing providers, here is how I would choose one, and here is what this work tends to cost.
How I put this together
On October 22, 2026, I read the three top-ranking US articles for "why does chatgpt recommend my competitor." Then I checked claims against OpenAI's and Google's own documentation and labeled every vendor study as a vendor study.
- Mersel names three signals: third-party consensus, content structure, and entity clarity . It is written for B2B brands, leans on a Bain finding that 85% of B2B buyers arrive with a "Day One List" , and says nothing about location, listings or Bing.
- Pomo gives four causes, including blocked crawlers and inconsistent facts. Its testing tip is good. But it says three of the four causes you can fix yourself within a week , a timeline I will not repeat, and it skips location entirely.
- The Answer Engine lists seven reasons with percentage bars, such as "Unclear Service Pages 85%" , with no stated source or method. I did not reuse them. Its claim that reviews do not register plays down sources that local data shows getting cited.
What this article adds: OpenAI's documented "near me" rewrite and location sharing with providers like Bing, the split between search and training bots, host and CDN blocking, labeled local data on citations and answer variance, a repeatable test, and no invented stats or timelines. For the bigger picture, see what AI search visibility is, generative engine optimization, GEO vs SEO and how to get found in ChatGPT. More posts are on the blog.
Frequently asked questions
Why does ChatGPT recommend my competitor when I have better reviews?
Reviews are only one input, and OpenAI has not published how it weighs them. Your competitor may be closer to the searcher, listed on the sites ChatGPT pulls from, or have a page that matches the question better. Review sites still matter: BrightLocal's vendor study found Yelp among the most cited local sources, with more use on ChatGPT.
Can I pay to be recommended by ChatGPT?
Not in any way OpenAI documents for site owners. Its guidance is to let OAI-SearchBot crawl your site, and its help center says placement is not guaranteed. Be wary of anyone selling a submission to ChatGPT or a guaranteed mention.
Does blocking GPTBot remove my business from ChatGPT search?
No. OpenAI says its GPTBot (training) and OAI-SearchBot (search) settings are independent. Blocking OAI-SearchBot is what keeps your site out of ChatGPT search answers , so that is the one to check.
Does ChatGPT use the searcher's location?
Yes. OpenAI says ChatGPT collects general location from the IP address and may share it with search providers. A 'near me' question can be rewritten into a search with the city in it. Users can also turn on device location, which is off until they enable it.
Do I need an llms.txt file or special schema to show up?
Google says no special AI text files or schema.org markup are needed for AI Overviews and AI Mode. Structured data is still fine if it matches what the page says. OpenAI's site owner guidance focuses on crawler access, not special files.
Why does ChatGPT give a different answer each time I ask?
The searches, location and sources can change from run to run. BrightLocal's vendor study of about 200,000 local AI searches found the list of businesses shifts between repeated runs, though businesses reappear more than half the time. Log repeated tests instead of trusting one answer.
How long until ChatGPT starts recommending me?
No one can honestly give you a date, and I won't. The one documented timing is that OpenAI says robots.txt changes take about 24 hours to reach its search systems. That is when the setting is read, not when you get named.
Sources
- OpenAI: Overview of OpenAI Crawlers
- OpenAI Help: Searching the web with ChatGPT
- OpenAI Help: ChatGPT search for Enterprise and Edu
- OpenAI Help: ChatGPT release notes (Location Sharing, March 26, 2026)
- Google Search Central: AI features and your website
- SearchOS: AI Overviews (summary of Google's guidance)
- The Keyword: Google explains what AI search optimization requires
- Search Engine Journal: Anthropic's Claude bots make robots.txt decisions more granular
- Momentic: List of AI search crawlers and user agents
- eSEOspace: ChatGPT search explained
- BrightLocal (vendor study): What 200k local AI searches tell us
- BrightLocal (vendor study): Top sources and directories for local AI search
- BCC Interactive: AI search statistics for local and small businesses (BrightLocal and SOCi figures)
- Aggarwal et al.: GEO: Generative Engine Optimization (arXiv 2311.09735)
- LeadRescue: summary of the GEO paper's findings
- Tim Soulo (Ahrefs) on X: vendor research on ChatGPT citations
- The Arising: Why does ChatGPT recommend my competitor instead of me?
- Mersel: Why ChatGPT recommends your competitor
- Pomo: Why ChatGPT recommends competitors
- The Answer Engine: Why ChatGPT recommends your competitors
