If your business does not appear in ChatGPT, the cause is usually mechanical, not mysterious: the model cannot reach your pages, cannot find them in the index it searches, or cannot see enough clear evidence to answer. That gap matters because roughly 800 million people use ChatGPT every week. When I look up a company that feels invisible, I check three things before touching the site: is it crawlable, is it indexed, and does a clear answer about it exist anywhere.
The scale of automated reading is easy to miss. Across one month on Vercel's network, GPTBot made 569 million fetches, and all AI crawlers combined made about 1.3 billion, close to 28% of Googlebot's 4.5 billion. Cloudflare measured GPTBot climbing from 2.2% to 7.7% of crawler traffic between May 2024 and May 2025, a 305% rise in raw requests.
The robots are visiting. The real question is what they can read once they arrive.
Source: Vercel.
What Should You Check First If ChatGPT Ignores You?
Start with the three checks that explain most cases, in order of speed. You can run all three in an afternoon without changing a single line of your website.
Ask ChatGPT with search on a buyer question in your category, and note which sources it cites.
Confirm your key pages return real text in plain HTML, not only after JavaScript runs in a browser.
Check that Bing has those pages indexed, since ChatGPT's browse path leans on that index.
If the answer names 3 competitors and never you, you have a visibility gap, not a ranking problem. That distinction decides everything you do next, so resist the urge to rewrite copy before you know which cause applies.
Which Of These Situations Describes You?
Most absent companies fall into a small set of recognizable situations. Find yours in the table, then follow the matching check and first action.
If you see this | Likely cause | What to check first |
|---|---|---|
ChatGPT says it has no information on you | Not crawled or not indexed | robots.txt and Bing index status |
It describes you with stale or wrong facts | Old or thin public evidence | homepage and about-page copy |
It recommends competitors, not you | Weak or missing proof | comparison and specification pages |
Your page ranks on Google but is blank to AI | JavaScript-only rendering | server-rendered HTML output |
Each row is a different job. Two of the four are pure access problems you can fix in a day, and the other two are evidence problems that take real editorial work.
Can ChatGPT Actually Reach Your Website?
Often the honest answer is no, and the reason is rendering. AI crawlers currently do not run JavaScript, so anything drawn only in the browser can read as a blank page to them.
In Vercel's month of data, ChatGPT's crawler fetched JavaScript files in 11.5% of requests and Claude in 23.84%, yet neither executed it. Broken paths make the problem worse. ChatGPT's crawler hit a 404 on 34.82% of fetches and chased redirects on another 14.36%, against Googlebot's 8.22%.
A crawler that spends a third of its visit on dead URLs collects far less of you. The deeper mechanics sit in AI Crawlers Don't Run Your JavaScript.
Source: Cloudflare.
Are You In The Index ChatGPT Reads?
ChatGPT's search does not read the open web live. It retrieves from a search index and then cites a few results, so a page missing from that index cannot be cited at all.
Seer Interactive checked 500 citations across 100 queries and found 87% of SearchGPT citations matched Bing's top organic results, versus 56% for Google. That sample is small, so read it as directional.
One vendor analysis of 548,534 pages reported ChatGPT cited only about 15% of the pages it retrieved. Being retrievable is necessary, and still not sufficient. Where those answers come from is unpacked in Where Do AI Citations Come From.
What Can You Actually Measure, And What Can't You?
You can observe more than people expect, and prove less than vendors imply. Keeping those two apart is the whole discipline.
You can observe which answers and sources appear for a fixed question on a fixed date, and whether your pages are crawlable and indexed. You can only infer why a model preferred a competitor, because the ranking is never disclosed. Traffic, leads, and revenue need your own analytics and CRM, not an AI-visibility tool reading tea leaves.
Whether an answer is fair or accurate still needs a human to judge it. No tool controls a ranking or guarantees a citation, and 1 public view count is not proof of a buyer. The limits of single-number dashboards are argued in There Is No Such Thing as an AI Visibility Score.
How Does Trovance Help You See Why You're Missing?
Trovance is built for exactly this diagnosis.
It runs a weekly sweep that observes how AI systems answer questions about your company: what they say, which sources they cite, and where they compare or recommend a competitor instead of you. Rather than collapse that into 1 score, it preserves the raw answers and the evidence behind them, so you can tell whether the gap is crawlability, a missing page, or thin proof.
From there it points at the specific asset that should exist, and helps you produce and publish the evidence a model can actually use. See how AI currently describes your company and which evidence may be missing or unclear, then decide what to fix first.
When you are ready, start your free trial with your own company.
The trial does not promise a citation or a ranking. It shows you the current answer and the sources shaping it, so your next edit aims at a real gap instead of a guess. That is the difference between reacting to 1 bad answer and running a repeatable check.
