Comprehension: being described wrongly
A distinct failure from being unknown, and a worse one, because the business is being actively recommended to the wrong people.
What comprehension means
Having reached and read a business, does an AI system describe it correctly? Its category, who it serves, where it operates, what it specialises in, and what it does not do.
An AI visibility evaluation separates this from retrieval because the two fail independently and have completely different fixes. Retrieval is usually a technical problem with a technical answer. Comprehension is almost always a clarity problem, and the clarity is usually missing because everyone inside the business already knows the answer.
Why it is worse than being unknown
An unknown business is absent from answers. A misunderstood business is present in the wrong ones. It gets recommended to people it cannot serve, who arrive, discover the mismatch, and leave. That produces wasted enquiries, poor conversion, and a slow accumulation of experiences that read to everyone involved as a marketing problem rather than a description problem.
The symptom to watch for: a rising number of enquiries that are obviously wrong-fit, with no change in marketing. That pattern is comprehension failure often enough to be worth checking before anything else.
The four ways comprehension breaks
Category ambiguity. The business describes what it believes rather than what it is. Aspirational positioning is legitimate marketing and it leaves a system with no reliable category to file the business under, so it picks one.
Contradiction across sources. The site says one thing, the business profile another, a directory a third. These do not average. They lower confidence in all three, and low confidence is why a business is skipped in favour of a clearer competitor.
Geographic vagueness. Serving a region without stating which, or claiming national coverage while every signal points at one city. Location is one of the strongest filters in a recommendation, and vagueness here removes a business from local answers entirely.
Undifferentiated language. Text that could describe any business in the sector gives a system nothing to distinguish it with. A page of well-written generic claims is a page that produces no specific understanding.
Why it is usually invisible from inside
Everyone in the business knows what it does, so nobody encounters the description the way a stranger does. Internal familiarity fills every gap automatically, and after a while it becomes impossible to read the site as someone who knows nothing. This is why comprehension has to be tested externally rather than reviewed internally, and it is why the finding so often surprises people.
What a good evaluation reports
Not a score for clarity. The actual description the systems produce, in their words, alongside what the business says about itself, with the specific contradictions named and located. A business can argue with that. It cannot do anything with a clarity rating of six out of ten.