It usually isn’t budget
The brand a model recommends is rarely the one that spent the most. It is the one the model can describe most confidently, and confidence comes from clarity and corroboration rather than volume. That is unusually good news for smaller companies, because both are earnable.
Before diagnosing, gather the evidence. Ask several assistants your category’s buying questions — who should I use for X, what should I look for, who are the leading providers — and record which companies appear and which sources are cited. That citation list is the diagnostic, and it distinguishes a competitor winning on their own pages from one winning on what others have written about them.
1. They answer the question you avoided
Pricing, timelines, what you don’t do, who you’re wrong for. Competitors who state these plainly become the source for every prompt that touches them, while your page full of tasteful ambiguity is unusable as a citation.
The pattern is easy to confirm. Open the competitor page that is being cited and look for the sentence containing a number, a threshold or a rule. It is almost always there, and it is almost always something your equivalent page discusses without committing to.
You don’t have to publish a full price list to win this — a defensible range with the variables that move it is enough to make the page quotable.
2. Their description of themselves never changes
One consistent sentence about what a company does, repeated across its site, its listings, and everywhere it is written about, is a strong signal. Five different clever variations are five weak ones. Marketing teams like variety; models reward repetition.
Audit yours in ten minutes. Write down how you describe your company on the homepage, in the footer, on your main directory profile, on your social bios, and in the boilerplate of your last announcement. Most companies find three or four materially different versions, sometimes claiming different categories entirely.
Consistency looks boring to the people writing it and authoritative to the systems reading it.
3. Someone else vouches for them
Third-party corroboration — being written about, listed, reviewed, cited — is what separates a claim from a fact in a model’s handling of it. Self-description sets the ceiling; outside sources determine how much of it is believed.
Not all corroboration is equal. A specialist publication describing precisely what you do is worth more than a general directory entry, and one detailed customer review naming the problem you solved is worth more than a dozen five-star ratings with no text — which is why review prose matters more than the average.
This is the slowest lever, which is exactly why it is the most defensible one. It cannot be bought in a sprint by a competitor who notices you’re winning.
4. They’re specific about a narrow thing
Models resolve narrow questions more confidently than broad ones, and narrow claims are easier to corroborate. A company known for one specific capability gets recommended for it; a company that lists eleven services equally gets recommended for none of them.
This is uncomfortable because it argues against how most service businesses present themselves. The resolution is not to stop offering things — it is to be unambiguous about what you are the obvious choice for, and to let the rest be discovered by people already talking to you.
The uncomfortable fifth reason
Sometimes the competitor is genuinely better known for genuinely good reasons: longer track record, more published work, more customers willing to talk about them. No content strategy substitutes for that in a quarter.
What does work is narrowing the ground you contest. You are unlikely to displace an established name on the broad category question, and you can absolutely own a specific one — a sector, a constraint, a particular kind of problem — where their generality becomes a disadvantage.
The order to fix them in
Answer the avoided questions first — it is a week of writing and it changes what you are eligible for. Fix your self-description second, because it is nearly free. Narrow your positioning third, because it requires a real decision. Build corroboration continuously, because it never finishes.
Doing them in the opposite order is the common failure: teams chase mentions while their own site still refuses to say what things cost. Corroboration amplifies whatever your site says about you, so amplifying an ambiguous description mostly produces ambiguity at greater volume.
What to expect, and when
The first two fixes change what you are eligible to be cited for and tend to show up within weeks. Narrowing shows up as better-fitting enquiries before it shows up in visibility. Corroboration moves over quarters, which is why the programme has to be judged on the right horizon or it gets cancelled just before it works.