The figures below come from AI Visibility Reports Avid ran during 2026. Brand names, categories, competitors and markets have been removed. These are diagnostic reads, not campaign results, and no result is implied by any number here.
Put a hundred questions to the same AI engines, in the same week, about the same brand. Word them so they point at what the brand already has, and the brand is named in 44% and 51% of answers across the two markets read. Word them the way a person actually asks, describing what they want and leaving the model to decide what matters, and the same brand is named in 11% and 13%.
A four-fold drop, in both markets, from nothing but the phrasing of the question.
Nothing about the brand changed. Nothing about the engines changed. The only thing that varied was whether the question started from the brand or from the buyer.
Some questions carry the brand already. Is this brand reliable? How do I set up this product? A model asked a question containing your name will nearly always say your name back to you.
Across three separate reads, the brand was named in 209 of 210 such answers, in 206 of 208, and in 282 of 299. The range we see runs from 94% to 99.5%, and it barely moves from brand to brand.
Average those in with everything else and you can publish a 95% AI visibility score. The arithmetic is honest. The number is close to meaningless, because most of what it measures is how many questions you wrote with your own name inside them.
This is why a report has to read each kind of question on its own measure instead of blending them into a single figure. The four kinds of question, and what each one is for.
In one read a brand was named at a rate close to the category leader, 26% against 30%, which ranked it second of six. On those same questions its average position inside the answer was a point and a half worse than the leader's.
A mention count cannot tell the difference between being named second and being named eighth. If measurement stops at whether you appeared, a loss reads as a draw.
Engines disagree, and they disagree more than most dashboards admit.
One blended figure hides all of that. It is worse than it looks, too: two of the three engines most reports track belong to the same vendor, so an average of three is nearer an average of one and a half.
If someone hands you an AI visibility number, ask these before you believe it.
A read that survives those four is worth acting on. One that does not is a number chosen to look good.
10 September 2026 · 6 min read
Before any engine can quote your title, the page has to clear conditions that have nothing to do with the writing. The six that decide it, the two halves of a quotable page, and the crawler distinction most robots.txt files have never made.
10 September 2026 · 6 min read
OpenAI's own documentation says a sponsored placement sits below the answer and cannot change it. That makes the ad and the recommendation two different purchases, out of two different budgets.
9 September 2026 · 6 min read
The largest kind of source in an AI answer is often one nobody can buy. Here is the subtraction that shows what is actually left to commission in your category.
See how your brand appears in AI answers, and where trusted publisher content can strengthen its visibility.
