Guide
Why doesn't ChatGPT recommend my business?
A customer messages you: “I asked ChatGPT and it didn’t say your name.” You open ChatGPT yourself, type the question your buyers actually ask, something like “best [category] for [use case]”, and watch it name three competitors in a row without mentioning you once. Citation is the act of an AI answer engine naming a specific business, page or source inside the answer it generates for a user. If that never happens for your business on a given engine, you simply aren’t cited there, and the reason is rarely that anything about the business is wrong. It’s a pattern in how the engine chooses who to name, and a pattern can be found and changed.
This piece exists because a pattern like that is checkable. Sourcive runs a company’s own buyer questions directly against the OpenAI, Anthropic and Perplexity APIs, which is what turns “ChatGPT didn’t say your name” from a suspicion into something you can actually look at.
Is this the same problem as ChatGPT getting your details wrong?
No, and the distinction matters because the fix is different. Getting your details wrong is a misdescription problem: the model has heard of you and says something inaccurate, outdated, or confuses you with another company of the same name. Why does ChatGPT describe my company incorrectly? covers that case in full. Omission is different. The model isn’t confused about who you are. It simply never retrieves or names you at all when a buyer asks the exact question that should surface you. You can be entirely accurate on your own site and entirely absent from the answer at the same time, which is why fixing your facts alone doesn’t solve this problem.
Why do ChatGPT and Perplexity recommend different businesses for the same search?
Because they aren’t running the same search. ChatGPT’s live answers draw on a Bing-indexed web search layer plus whatever the underlying model already learned in training. Perplexity runs its own crawler, PerplexityBot, against its own index, and shows sources as a numbered list beside the answer rather than folding them into prose.
| ChatGPT | Perplexity | |
|---|---|---|
| Retrieval | Bing-indexed search plus training data | Own crawler and index (PerplexityBot) |
| Citations shown as | Named inline within a synthesised answer | Numbered source list beside the answer |
| Live lookup | Triggered when the model judges it’s needed | Runs on effectively every query |
Two systems, built by two companies, crawling on two schedules, ranking relevance by two different methods, won’t converge on the same shortlist just because the words in a query match. A business cited on one has no guarantee of even being retrieved by the other. Treating “AI SEO” as one target, fixed once, treats these as one system. They aren’t. A fix aimed at only one of them will look like it failed everywhere else on its own.
Does it matter where on the page the answer sits?
Yes, and it’s a structural point rather than a ranking one. Answer engines assemble a response from passages, not whole pages, and the passage that most directly answers the query is the one likeliest to get pulled in and named. A homepage that opens with a mission statement, a hero image caption and two paragraphs of scene-setting before it says what the company actually does is asking the retrieval step to do extra work to find the one sentence worth citing. Say what you do, who it’s for and where, in plain sentences, near the top. A file called llms.txt, sitting at the root of a site as a plain-text signpost to the pages worth reading, exists for exactly this reason: it tells a crawler where the load-bearing content is instead of leaving it to guess.
How do you actually find out where your business stands?
You measure it, engine by engine, rather than guessing from one ChatGPT conversation on one afternoon. That means running your own buyer-intent questions against the same OpenAI, Anthropic and Perplexity APIs the engines use, and recording whether and how your business gets mentioned in each one, so you end up with a per-engine pattern rather than one blended score. A business cited by Perplexity and invisible on ChatGPT needs a different fix than a business invisible on both. Sourcive takes a company’s URL, works out what it does and who it competes with, and runs exactly that measurement, engine by engine, rather than guessing at it. That’s the finding worth having before rewriting a single page.
Common questions
Does fixing my homepage guarantee ChatGPT will start citing me? No. It makes citation possible where it currently isn’t, by giving the retrieval step a clean, direct passage to pull from. Whether a specific engine actually surfaces you on a specific query still depends on everything else it retrieves for that query that day, which is why checking per engine matters more than any single fix.
Is this just SEO with a new name? Related, not identical. Classic SEO ranks pages against a search engine’s link graph. AI citation depends on which passages a model’s retrieval step pulls and judges worth naming inside a generated answer. A page can rank first on Google and still never get cited by ChatGPT or Perplexity, because the two systems are answering different questions about the same page.
Can a tool actually tell me which buyer questions I’m missing? Yes. A tool that runs your own buyer-intent questions against the same APIs the engines use, and records which ones you never appear for, turns a suspicion into a checkable list instead of a general instinct that something’s wrong. Sourcive’s paid report is built around exactly that list, alongside the pages and communities the engines cited instead of you. Guessing is free. Finding out isn’t, and neither is losing the customer who already asked and got someone else’s name back.
Sourcive measures whether ChatGPT, Claude, Perplexity and Gemini mention your business when your buyers ask, and gives you the files to fix it. Run the free audit.