Positioning and visibility
Why AI thinks you compete with companies you've never heard of
By Arnav Mukherjee, founder of TofuBofu · July 16, 2026
A few days ago I ran our own product on ourselves. TofuBofu, scanning TofuBofu. The score came back a flat zero: not named once across ten buyer questions on ChatGPT, Claude, Perplexity, and Gemini. We are a young brand, so I expected the visibility to be thin. What I did not expect was who the AI put in our place.
Ask those engines for the best tools that do what we do, and they do not reach for a rival like us. They default to the SEO tools they already trust: SEMrush, Ahrefs, Moz. Then it got stranger. When I asked the engines what TofuBofu actually is, they could not agree on a category at all. One described a platform for selling online courses. Another called us a content marketing agency. ChatGPT read the name literally and decided we optimize the top and bottom of a sales funnel, the exact opposite of what we do. The machine had no stable idea what we are, so it invented a few and pointed buyers at the incumbents. That confusion turned out to be the single most useful thing the scan told me.
The competitor set is an answer, not a distraction
When a scan shows you the companies an AI engine names in your category, the instinct is to skim past the ones you do not recognize as noise. Do not. That list is a map of how the machine has filed you. It is the closest thing you can get to reading your own positioning through a stranger's eyes, except the stranger is the system your buyers are now asking for recommendations.
There are two readings, and both are information. If the engine names your real rivals, the ones you meet on shortlists and in sales calls, your category is legible: the machine knows what you are, which is the precondition for it recommending you. If the engine names companies from an adjacent market you do not actually compete in, your content is broadcasting the wrong signal, and the list just told you which wrong signal.
This is why we treat the mismatch as a headline finding in a report, not a footnote. You think you compete with A and B. Your content makes AI frame you against X and Y. Your buyers, asking the same engines, get that same wrong frame. The gap between the two sets is a defect you can actually fix, and most of your market has not noticed theirs.
Why the machine miscategorizes you
An engine does not know your business model. It knows the words that appear on and around your site, and it groups companies by the company they keep in text. Two brands that get described with the same vocabulary, and that turn up together on the same listicles and comparison pages, get clustered together, whether or not they actually compete. The competitor set you see is that clustering, read back to you.
So the miscategorization has causes you can name. The first is thin content that never states your category in plain words, which leaves the model to guess from context. The second, and the one that got us, is borrowed vocabulary, and our own name is the trap: TofuBofu reads as funnel-marketing jargon, ToFu and BoFu, so an engine going literal files us as a funnel tool before it reads a word of what we actually do. The third is weak corroboration: if few independent sources place you in your real market, the engine has little to counterweight the guess.
What the mismatch costs you
This is not a cosmetic problem. According to G2's 2026 buyer research, 51% of B2B buyers now start vendor research on an AI chatbot, up from 29% a year earlier, and 69% say they switched their choice of vendor based on what AI told them. The frame the engine builds is the frame your buyer inherits before they ever reach your site.
A wrong competitor set costs you in two directions. When AI files you in the wrong category, you simply do not appear for the category queries you should win, because the model does not think you belong there. And when you do appear, you can surface next to companies you do not actually compete with, on criteria that do not favor you, so you lose the comparison before it starts. Either way the buyer is quietly steered, and you never see the deal you did not get.
See which competitors AI names in your place
A free scan runs your buyers' questions across six AI engines and shows you the exact companies each one lists, so you can spot the mismatch yourself.
Run your free scanHow to fix the frame
You cannot tell the model which list to print. You can change the evidence it builds the list from. Every move below feeds the same goal: being read as a member of the category you actually sell in.
1. State your category in plain words, repeatedly
Your homepage, about page, and product pages should say what you are in a sentence a stranger could repeat: the category noun, who it is for, what job it does. Not the aspirational metaphor, the plain label. If a human skimming your site cannot name your category in five seconds, neither can a model, and it will borrow one.
2. Use the vocabulary of the rivals you want
The words you choose decide the cluster you join. If you want to sit beside the real alternatives in your market, use the language those alternatives and their buyers use, and state your category in plain words the model cannot misread. Our own name reads as funnel jargon, so we have to hard-anchor the phrase answer engine optimization everywhere, or the engine keeps inventing a funnel tool. Say plainly what you are, and the frame tightens.
3. Publish comparison pages against your real competitors
A clear 'you versus your actual rival' page does two jobs. It captures buyers evaluating that specific alternative, and it hands the engine an explicit, structured statement that you and that company belong to the same category. Write these against the rivals you want to be measured against, not the ones the AI mistakenly chose.
4. Build third-party corroboration in your category
Reviews on the platforms your category lives on, mentions in roundups and listicles for your market, participation in the communities where your buyers gather. Each independent source that files you in the right category outweighs a guess. This is the same off-site work that drives visibility generally, aimed at a specific target: category membership.
5. Re-scan and watch the set migrate
The competitor set is your scoreboard for this work. Re-scan monthly. Search-grounded engines will shift first as they re-read your updated site and profiles, then the wrong-category names should thin out and the right ones fill in. When the list AI prints starts matching the list in your head, the machine finally understands what you are.
The part I did not expect to be grateful for
Scoring zero on our own product was not fun. But the wrong competitor list was worth more than a clean score would have been, because it named the exact defect: our words read as a category we do not sell in. A friendly number would have hidden that. This is the reframe I keep returning to. The competitor set is the rare metric that tells you not just that you are losing, but why, in the precise vocabulary of the fix. Read it as a diagnosis, not an insult, and it becomes the shortest path to being framed against the companies you actually beat.
Frequently asked questions
What does it mean when AI names the wrong competitors for my company?
It means the AI has clustered your company with the wrong set of businesses based on the words on your site and around the web. Engines group companies by how often they appear together and by shared vocabulary, not by your real market. If your content does not clearly anchor you to your category, the model borrows a nearby category that your language happens to overlap with. The wrong competitor set is a direct readout of that miscategorization.
Why does AI put my company in the wrong category?
Three common reasons. First, thin or generic content that never states your category in plain words, so the model guesses. Second, borrowed vocabulary: you describe yourself using the buzzwords of an adjacent category, and the model takes you at your word. Third, weak third-party corroboration, so the engine has few independent sources placing you in your actual market. Fix any of the three and the frame starts to move.
Is the competitor set AI shows me useful or just noise?
It is one of the most useful things a scan gives you. The list of companies AI names in your place is a map of how the machine understands your positioning. If it names your real rivals, your category is clear. If it names companies from a different market, your content is sending the wrong signal, and you now know exactly which one. Treat the mismatch as a headline finding, not a data-quality complaint.
How do I get AI to compare me against the right competitors?
Name your category explicitly and repeatedly in plain language, use the vocabulary of the alternatives you actually want to sit beside, publish comparison pages against your real rivals, and build third-party mentions on the platforms your category lives on. Each of those gives the engine more evidence for the cluster you belong to. Then re-scan and check that the named set shifts toward your real market.
Can I control which companies AI lists alongside mine?
You cannot dictate the list, but you strongly influence it. The competitor set is generated from the text the engine has read about you and the companies you co-occur with. Change that text, on your own site and on the third-party pages that mention you, and the clustering changes. It is the same corroboration work that drives visibility, pointed at a specific goal: being read as a member of the right category.
How fast does the competitor set change after I fix my content?
Search-grounded engines like Perplexity, Copilot, and Google AI Overviews re-read the live web, so category-clarifying changes to your site and directories can move their framing within weeks. Training-based knowledge inside ChatGPT and Claude updates on model release cycles, so those shifts take longer. Corroboration on third-party sources compounds over months. Re-scan monthly to watch the set migrate.
Why would AI name a category I do not even sell in?
Because your words, and sometimes your own name, can read as a category you are not in. If the language on your site overlaps with an adjacent market, or your brand name sounds like something else, an engine going literal files you there before it verifies your business model. It is pattern-matching vocabulary, not checking what you do. Specific, plainly stated category language is what pulls you out of the wrong bucket.
Sources and further reading
- G2 2026 AI Search Insight Report: 51% of B2B buyers start research on an AI chatbot and 69% switched vendor based on AI, which is why the frame the engine builds matters
- Profound research on AI answer engines: how brands cited together and across more platforms compound their presence, the co-occurrence effect behind clustering
- Lewis et al., Retrieval-Augmented Generation (arXiv:2005.11401): the mechanism by which engines ground answers in the documents they read, the same documents that decide your category