Positioning and visibility
AI compares you to the wrong competitors. Do you write negative or positive content to fix it?
By Arnav Mukherjee, founder of TofuBofu · July 22, 2026
On a demo recently, a founder watched our scan show her the companies AI names when buyers ask about her category, and half of them were firms she did not consider competitors at all. Different market, different buyer. She got it immediately, this is a positioning problem, not a data glitch, and then she asked the sharpest question I have heard on the subject: "So do I write negative content, blog posts explaining these aren't really our competitors? Or positive content, here's who we actually are and why we're better?"
It is the right question, because the two approaches feel similar and pull in opposite directions. One of them mostly backfires. The other is the actual fix. The difference comes down to how AI decides which companies belong together in the first place, so let me start there.
The mismatch is a diagnosis. This is the prescription.
We have written before about why AI thinks you compete with companies you have never heard of: engines cluster businesses by the company they keep in text. Two brands described with the same vocabulary, appearing together on the same listicles and comparison pages, get grouped together, whether or not they actually compete. The competitor set AI shows you is that clustering, read back. That post diagnoses the problem. This one answers what to do about it.
And the mechanism is the whole answer. If AI groups you by co-occurrence, then every choice about content is really a choice about which names you want to appear next to. That single idea decides negative versus positive.
Why the negative approach usually backfires
The instinct is to write the rebuttal: "Why we are not like [wrong competitor]." It feels like setting the record straight. But think about what that page actually is to a machine reading it: a document with your name and the wrong competitor's name, together, repeatedly, discussing them as related enough to compare. You have just published fresh co-occurrence signal for exactly the association you want to break.
The engine does not read your argument and conclude "ah, they are different." It reads two names in one document about a shared topic and strengthens the link between them. Denial and association look the same in the data. This is the classic trap: you can spend a whole content calendar loudly distancing yourself from the wrong crowd and, in doing so, cement your place in it. Negative refutation content is the content equivalent of repeating a rumor to deny it.
The positive approach: build the right cluster
The fix is not to argue about the wrong category. It is to overwhelm it with signal for the right one. You want the machine to see your name surrounded by the companies, vocabulary, and topics of the category you actually belong to, until that association is stronger than the accidental one. Three things build it:
1. State your real category, in plain and repeated language
Your homepage, product pages, and about page should name your category in words a machine cannot misread, and use the vocabulary your real buyers and real alternatives use. This is the single biggest lever, because it is the text the engine reads most about you.
2. Publish comparison pages against your ACTUAL competitors
A clear 'you versus your real rival' page hands the engine an explicit, structured statement that you and that company belong to the same category. This is the deliberate, useful cousin of the negative page: same mechanic, but now you are co-occurring with the right names, on purpose, and capturing real comparison intent at the same time.
3. Earn third-party mentions in your true category
Reviews on the platforms your real category lives on, inclusion in roundups and directories for your actual market, participation in the communities where your true buyers gather. Each independent source that files you correctly outweighs the accidental signal, and third-party corroboration is what AI leans on hardest.
See which competitors AI names in your place
A free scan shows the exact companies each AI engine lists for your category, so you can tell the accidental associations from the real ones and fix the frame.
Run your free scanThe one time to name the wrong comparison
There is a real exception, and it is where the founder's instinct was half right. Sometimes buyers genuinely consider a company you see as unrelated, they type "your category vs that company" into AI, even though you think the comparison is silly. When the query is real, capturing it beats ignoring it. But write it as a reframe, not a rebuttal: acknowledge the overlap honestly, help the buyer understand when each fits, and draw the line that puts you in your own category. That page captures real intent and reclassifies you at the same time. The failure mode to avoid is the purely defensive "we are nothing like them" post that captures no intent and only deepens the wrong link. The test is simple: are you publishing because a buyer searches for this comparison, or only because it annoys you? The first is strategy. The second is the trap.
Frequently asked questions
Should I write content saying a company is not my competitor?
Usually no. A page that argues 'X is not our competitor' still puts your name and X's name in the same document, which is exactly the co-occurrence signal that taught the AI to associate you in the first place. You reinforce the link you are trying to break. The exception is a genuine comparison page a buyer would actually search for, where you use the page to reclassify yourself, not just to deny the association.
How do I fix AI comparing me to the wrong competitors?
Build positive signal for the category you do belong to, rather than negative signal against the one you do not. State your category plainly and repeatedly on your site, use the vocabulary of your real alternatives, publish comparison pages against the rivals you actually compete with, and earn third-party mentions in your true category. AI clusters companies by the company they keep in text, so the fix is to be seen with the right names, not to argue about the wrong ones.
Why does AI put me in the wrong competitive set at all?
Because it groups companies by shared vocabulary and by which names appear together in the content it reads, not by your real market. If your site uses the buzzwords of an adjacent category, or your brand name reads like something else, or you have thin third-party corroboration in your actual niche, the model files you where your language points. The wrong competitor set is a direct readout of that miscategorization.
Is it ever right to publish a comparison page against a company I don't compete with?
Yes, when buyers genuinely consider that company as an alternative to you, even if you see the two as different. Then the comparison query is real and worth capturing, and you use the page to reframe: acknowledge the overlap honestly, then draw the line that puts you in your own category. That is different from a defensive 'we are nothing like them' post, which only deepens the wrong association without capturing any intent.
How long does it take to fix a competitor mismatch?
Search-grounded engines like Perplexity and Google AI Overviews re-read the live web, so category-clarifying changes to your site and profiles can move their framing within weeks. Training-based memory inside ChatGPT and Claude updates on model release cycles, so those shifts take longer. Re-scan monthly and watch the named competitor set migrate toward your real market as the new signal accumulates.
Does the competitor mismatch actually cost me deals?
Yes, in two ways. When AI files you in the wrong category, you 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 be compared on criteria that do not favor you, against companies you would never lose to in a real evaluation. 69 percent of buyers say they switched vendor based on AI guidance, so the frame the engine builds is the frame the buyer inherits.
Sources and further reading
- Profound research on AI answer engines: how brands cited together and across more platforms compound their presence, the co-occurrence effect that drives clustering
- G2 2026 AI Search Insight Report: 69% of B2B buyers switched vendor based on AI guidance, so the competitive frame AI builds is the one the buyer inherits
- 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 which category you are grouped into