You open ChatGPT, you ask it what it recommends for a product in your category, and it comes back with three names but not one of them is yours. It is one of the most common frustrations in B2B right now, and it can cost you deals. G2 found that 51 percent of software buyers now begin their research with an AI chatbot more often than with Google, up from 29 percent a year earlier, and that one in three of them went on to buy from a vendor they were not even familiar with beforehand (G2, The Answer Economy, 2026). With ChatGPT alone at 900 million weekly users, whoever the model names gets a shot at the deal, and whoever it leaves out never enters the room.
You are probably not invisible
If your site ranks on Google, ChatGPT has almost certainly encountered you somewhere, and the idea that you are completely invisible to it is usually wrong. I have seen plenty of "I rank number one but I do not exist on ChatGPT" posts, and most of them are describing something narrower than true invisibility.
You are not missing from the model's memory, you are missing from its answer, and only for the questions that matter. The question is not why ChatGPT has never heard of you, it is why it does not put you in the answer when a buyer asks who they should actually use, which is a much smaller and much more solvable gap. The same holds for Gemini, Perplexity, and Claude, and I use ChatGPT throughout only because it is the one people name most, not because the reasons are unique to it.
The problem is that being known is not the same as being recommended.
It never learned enough about you
A model's base knowledge comes from a training snapshot of the web, and it holds the entities that show up often across the sources it ingested. If you are newer, smaller, or written about mostly on your own site, you are thin in that memory, and thin things do not make it into a confident answer. This is the part the "ranked but never cited" crowd gets half right: a strong search profile and a strong AI-memory profile are built from different materials.
The signal that matters here is not your backlink count or your domain rating, it is whether independent, editorial sources the model trusts actually talk about you, in their own words, rather than just linking to you. If your presence is a wall of links but almost no one describes what you do, the model has nothing solid to repeat. This is a slower lever than the rest, and it overlaps with the broader shift I wrote about in GEO, the new SEO.
Can ChatGPT even read your site when it checks?
When ChatGPT does go and look at your site, it sends a crawler, and if that crawler cannot get in or cannot parse what it finds, you are skipped. The usual culprits are GPTBot or OAI-SearchBot blocked in robots.txt or by a firewall, a site that only renders in JavaScript or pages where your real facts live inside design rather than in clean, readable text.
This one is worth checking first, because it is the most fixable and the most common. If the crawlers cannot reach you, nothing else you do to your content matters, since the model never sees it. I walked through what each of these bots actually fetches in what AI bots are doing on your site.
It cannot tell who you are
If your name collides with another company, a common word, or a well-known person, the model hedges or picks the more established meaning, and you lose the mention to your own namesake. The same thing happens when your positioning is vague, because a line like "the platform for modern teams" gives the model nothing concrete to attach to you. Models reward clarity and consistency, a single, stable description of what you are that reads the same everywhere they find you.
Why do your competitors own the answer?
For "best tool for X" and "alternatives to Y" prompts, the model reaches for the names that dominate the conversation it learned from, and if three competitors are mentioned everywhere while you are mentioned rarely, they fill the shortlist and you do not. This is where being ranked but not recommended bites hardest, because you can sit on page one of Google and still lose the consideration set inside the answer, and G2 now puts AI chatbots as the single most influential source shaping which vendors make a buyer's shortlist at all.
It knows an old version of you
Sometimes the model is not wrong about you so much as out of date, working from an old pricing page, a two-year-old forum thread, or a rebrand it never caught. When that happens it either describes a you that no longer exists, or it leaves you out because what it has is too thin and too stale to be worth putting in the answer. Freshness is one of the strongest levers you have, and it is closely related to why AI gets other brand facts wrong, which I covered in what do_not_infer does.
How do you tell which one is your problem?
Each cause leaves a different fingerprint so you can start by running the prompts a buyer would actually type, not just "what is my company" but "best tool for this job" and "alternatives to this competitor", across ChatGPT, Claude, and Perplexity, and note whether you appear and in what company. There is a fuller version of that audit in the five prompts to ask AI about your brand.
Then check your server logs for GPTBot, OAI-SearchBot, ClaudeBot, and PerplexityBot, because if they are not visiting at all, then live retrieval is not what is failing you, and the gap is in what the model remembers, which points you back at footprint and freshness. Finally, load your key pages the way a bot does, with JavaScript turned off, and see whether your core facts still survive on the page. Between those three checks you can usually tell within an afternoon which of the causes above is actually yours.
What actually moves the needle?
You cannot reach into a model and order it to recommend you, and anyone promising a guaranteed mention is selling you something that does not exist. What you can do is remove the reasons it leaves you out, and two of those reasons, an unreadable site and unclear facts, are directly in your control, which is the part we work on at Ooky.
Brand Intelligence compiles a clean, structured profile of your brand and serves it to the AI crawlers at the edge, so that when GPTBot or PerplexityBot comes to look, it finds accurate, machine-readable facts instead of an empty shell or a guess, and for the fields you cannot afford to have invented, a do_not_infer directive tells the model to leave them out rather than fill the gap. Brand Tracker then runs those buying prompts on a schedule across models and personas, so you can watch whether the answer actually changes over time instead of checking once and hoping.
I will not tell you this makes ChatGPT recommend you, because no tool can truthfully promise that. What it does is give the model something accurate to find, and give you a way to measure the movement, which is the difference between guessing and knowing. The free tier covers the tracking, so you can start by seeing whether ChatGPT, Gemini, Perplexity, and Claude mention you at all before you spend anything on fixing it.
FAQ
If I rank number one on Google, why am I not in ChatGPT's answers?
They overlap more than people assume, but they are not the same thing. A strong Google ranking usually does mean the model has encountered you, and that presence is the groundwork. Landing in the answer also depends on whether the model trusts the sources that describe you, whether its crawler can read your site, and whether you clearly match the question, so ranking helps but does not guarantee you a spot on the shortlist inside the answer.
Does ChatGPT use my website, or only its training data?
Both. Its base knowledge comes from a training snapshot with a cutoff, which changes slowly, and when browsing or retrieval is on it also fetches live pages at the moment of the question. Retrieval is the faster lever, because a re-crawl can change the answer long before the next model is trained.
How do I get ChatGPT to mention my company?
You cannot force a mention, and anyone who promises one is selling you something. What you can do is remove the reasons it leaves you out: let its crawlers reach clean, machine-readable facts about you, build genuine third-party presence in the sources it trusts, and keep your positioning clear and consistent so the model knows exactly who you are.
How long does it take to show up in ChatGPT?
It depends on the lever. Retrieval-based changes can appear as soon as the crawler re-visits and the answer is generated again, sometimes within days, while training-based memory moves on the model's own schedule and is much slower. Keeping your facts fresh and crawlable is what you can control.
See why the model leaves you out, and fix what you can
Ooky serves AI crawlers a clean, structured version of your brand, then tracks whether ChatGPT, Claude, Gemini, and Perplexity actually start mentioning you. The free tier covers the perception tracking, so you can start by seeing the gap clearly before you spend anything.
Sources
- G2. "The Answer Economy: How AI Search Is Rewiring B2B Software Buying." Survey of 1,076 B2B decision makers, March 2026. Retrieved 2026-07-15. https://www.prnewswire.com/news-releases/new-g2-research-half-of-b2b-software-buyers-now-start-their-research-with-ai-chatbots-302742807.html
- TechCrunch. "ChatGPT reaches 900M weekly active users." February 27, 2026. Retrieved 2026-07-15. https://techcrunch.com/2026/02/27/chatgpt-reaches-900m-weekly-active-users/
- OpenAI. "Overview of OpenAI crawlers (GPTBot, OAI-SearchBot)." Retrieved 2026-07-15. https://platform.openai.com/docs/bots