Why Isn't ChatGPT Mentioning My Business? Seven Checks

The question usually arrives as a screenshot. Someone asked ChatGPT for the best remodeler, clinic or software vendor in their area, got a confident paragraph naming three competitors, and their own business was not in it. The instinct is to treat this as a content problem. In our experience it is more often a plumbing problem, and plumbing is cheaper to fix.

Work the checks in order. Each one is a precondition for the next, and there is no point rewriting your service pages if the agent that would read them is being served a 403.

Check 1: Is OAI-SearchBot allowed to fetch your pages?

OpenAI runs several agents and documents them separately. The one that matters here is OAI-SearchBot, which OpenAI says “is used to surface websites in search results in ChatGPT’s search features”. The same page is blunt about the consequence of blocking it: “Sites that are opted out of OAI-SearchBot will not be shown in ChatGPT search answers,” although they can still appear as navigational links.

Two things to look for. First, your robots.txt. A blanket User-agent: * with Disallow: /, or a well-meaning block aimed at “AI bots” that caught the search agent along with the training crawler, is enough. OpenAI documents GPTBot as the training crawler and says each setting is “independent of the others,” so you can disallow GPTBot and still allow OAI-SearchBot if that is the policy you want. Our robots.txt generator writes the per-agent rules for you.

Second, your CDN or firewall. Bot-management rules can return a 403 or a challenge page to anything that is not a mainstream browser, and robots.txt never gets a say. That failure mode is common enough that it has its own ten-minute check. The quick version: fetch a page with the bot’s user-agent string and compare the status code to a normal request. OpenAI notes that robots.txt changes take about 24 hours to be picked up, so do not re-test the same afternoon and conclude nothing happened. A fuller list of who crawls what is in which AI crawlers visit your site.

Check 2: Is the page indexed anywhere at all?

Assistants retrieve from indexes. A page that no index holds cannot be returned for any question, however relevant it is. This is the single most common cause we find on sites that “publish constantly” and still get ignored: the pages exist, and search does not know they do.

Open Search Console, inspect the URL you expected to be cited, and then look at the Pages report for the ratio of submitted to indexed URLs across the whole site. If a large share sits in “Discovered, currently not indexed” or “Crawled, currently not indexed,” you have an index coverage problem, and adding pages will make it worse rather than better. Google’s own guidance for its AI features says the same thing from the other direction: a page must be indexed and eligible for a snippet before it can appear in AI Overviews or AI Mode, with “no additional technical requirements” beyond that.

Check 3: Does the content exist in the server-rendered HTML?

Open the page, use View Source rather than Inspect Element, and search for a full sentence from the body copy. If the sentence is not there, the page is being assembled in the browser by JavaScript, and you are relying on every retrieval agent to execute that JavaScript before reading. Google documents rendering for its own crawler. The OpenAI, Anthropic and Perplexity documentation we link in this article does not describe JavaScript rendering at all, and the safe engineering assumption is that a fetch returns what the server sends.

The fix is server-side rendering or static generation for every page you want quoted. It is one of the reasons we build client sites the way we do, and it is covered in our AI search optimization service page.

Check 4: Can a stranger state who you are in one sentence?

A model answering “who does X in Y” needs four facts about you: the name, the service, the place, and the kind of customer. Many sites bury all four. The homepage leads with a tagline (“Spaces that inspire”), the About page tells the founder’s story, and the service area is implied by a phone area code. Humans fill in the gaps. A retrieval system is not paid to.

Read your homepage and About page as if you had never heard of the company. If you cannot write “[Name] is a [service] company in [place] that works with [customers]” from what is on the page, neither can anything else. Put that sentence on the page, make the name identical everywhere it appears, and back it with Organization and LocalBusiness structured data that says the same thing. This is the practical core of entity SEO; if the term is new, start with what an entity is.

Check 5: Is there a sentence worth quoting?

Assistants compose answers from pieces of pages. The pieces they prefer are self-contained: a definition, a numbered procedure, a price, a comparison, a direct answer to a direct question. A page that is all narrative, however well written, gives them nothing to lift cleanly.

Take your five most common customer questions and find the exact sentence on your site that answers each one. If the answer is spread across three paragraphs, or only lives in a brochure PDF, or is in an image of a price list, rewrite it as a sentence a stranger could quote without context. Keep the narrative; add the answer above it. Where the same question and answer is shown on the page, FAQ structured data is appropriate; where it is not, leave the schema off.

Check 6: Does anyone other than you say you exist?

A business described only on its own website is a claim. The same business described on a Google Business Profile, in two or three industry directories, in a local news story, and in reviews that mention the service and the place, is a fact that several independent sources agree on. Retrieval systems lean on that agreement because it is harder to fake than a web page.

Search your business name in quotes and read what comes back. Then check the basics: is the name, address and phone number identical across your site, your Google Business Profile and the directories that list you? Are there reviews, and do they mention what you do and where? Is there any press, a chamber listing, a supplier page, a client case study that names you? Each of these is a sentence somewhere else on the web that corroborates a sentence on your site.

Check 7: Have you measured it, or assumed it?

One screenshot from one account is an anecdote. Assistants vary their answers by phrasing, by location signals, by whether search is enabled, and from day to day. Before deciding the problem is “never,” make it a measurement.

1. Write 10 to 20 questions your customers actually ask, in their words, including the place where it matters.
2. Run each one in ChatGPT with search enabled. Record every business and source that is named, and whether yours is among them.
3. Repeat on a fixed schedule, monthly at least, from the same account and settings.
4. Change one thing on the site between runs, so a movement can be attributed to something.

This is boring, and it is the only way to know whether checks 1 to 6 did anything. Pair it with the referral traffic you can see in analytics from chatgpt.com and the other assistant domains, and with the server-log evidence of which agents actually fetched which pages.

The seven checks in one table

What not to do

• Do not block every bot in a panic, and do not unblock every bot in a panic. The search agents and the training crawlers are documented separately by OpenAI, Anthropic and Perplexity, and the decisions are separate. Allow the agents that cite; decide on the training crawlers as a policy question.
• Do not stuff “ChatGPT” into your copy. A page about plumbing that mentions AI assistants six times is not more likely to be quoted about plumbing. Clarity about what you do is the signal.
• Do not buy mentions. Paid directory spam and fabricated reviews are the opposite of corroboration: they are a second claim from the same interested party, and they are the kind of pattern these systems are built to discount.
• Do not skip the index check because the site “ranks fine.” Ranking for the brand name says nothing about whether the service pages that would answer a customer question are indexed.

Frequently asked questions

About the work

Webvello is an AI search optimization and index engineering company founded in 2024, based in Roseville, California. We have delivered 50+ projects across 37+ cities, with a 3–5x average client ROI on properties where the technical foundation was the binding constraint. Founder George Shvaya came to search from cybersecurity and systems engineering, which is why this article starts with access logs rather than adjectives. The full method, organised by stage, is in the AI Search Playbook.