Being mentioned is not being recommended

There is a measurement everyone in this field reports and a second measurement almost nobody does, and the gap between them is where most of the money is wasted.

The first is how often you get named. Ask a hundred buying questions, count the answers that contain your name, divide. Every tool sells this number.

The second is what happens when you are named. Being listed eleventh in a list of twelve is a mention. Being the one the model says to call is also a mention. The tools record both the same way, and they are worth wildly different amounts.

A June 2026 visibility analysis published a good example of the distinction, and then drew a conclusion from it that the numbers do not support. Both halves of that are instructive, so let’s do both.

The example

Two companies in the same category, measured across a thousand AI responses:

Apollo      645 mentions      21 in every 100 were a genuine recommendation
Lavender     59 mentions      32 in every 100 were a genuine recommendation

The published reading: the smaller company counts for more every time it is named, because it says more precisely what it does. Fewer mentions, more weight.

The idea is right. The evidence for it, at these numbers, is not there.

Why the numbers don’t carry the claim

Fifty-nine cases is not many, and a proportion worked out from a handful of cases can hide very different underlying values. Here is the range each number could reasonably sit in:

Lavender    32 in 100, counted on 59 cases      the true figure is somewhere from 20 to 44
Apollo      21 in 100, counted on 645 cases     the true figure is somewhere from 18 to 24

Those two ranges nearly touch. The smaller company’s true figure could be 20, which is the same as the larger one’s. Or it could be 44, which would make the headline an understatement. That measurement cannot tell you which, and publishing the bare number without its range makes the doubt disappear.

This is not a nitpick and it is not a gotcha. It is the difference between a number you can act on and a number that reads well. If you rebuild how you present your company because 32 beat 21, you have made a decision by looking at a coin flipped fifty-nine times.

What would have made it hold? More attempts. The same hundred questions asked five times each instead of once takes the smaller company from 59 mentions toward 300, and the uncertainty narrows from twenty-four points to about five. Then the comparison means something. It costs a few hours of computing time.

Note what that study’s own methodology page does not say: how many times each question was asked. It says ten models, a hundred questions per category. It never says how many times each question was repeated. On a system that answers the same question differently twice in a row, asking once is not measuring. It is looking at a single case, a hundred times over.

The distinction is still the right one

Now the useful part, because the flawed example was pointing at something real.

Two businesses named equally often can be in completely different commercial positions, and the cure for each is different.

Named often, but almost never recommended. The models know you exist. They list you. They send nobody to you. You are furniture in the category, named because you are one of the names, not because you fit the question. This is a positioning problem, and publishing more about yourself makes it worse, not better: more generic material about a generic company produces more generic mentions.

Named rarely, but recommended when you are. When you come up, you win. You just rarely come up. This is an awareness problem, and it is the easier of the two to fix. The thing that works is being present in more of the places the models read, so that the questions where you already fit actually surface you. Written by somebody else has where those places are.

Low on both. Start at gate one. Something more basic is wrong.

High on both. Defend it. You are in the spot everyone else in your category is paying to reach, and those spots change hands more often than they look.

How to tell which one you are, honestly

Three rules, and they are the whole method.

Ask the question more than once. The same prompt to the same model on the same day produces different names. If your report asked each question once, it photographed pure chance. We ask twice in the first pass, at separate moments, and a third time on day seven, and we publish how far apart the answers were, not just the average.

Separate the count from the outcome. Record, for every appearance: were you named, and were you the answer? Those are two columns, not one. Any report that gives you a single “visibility score” without showing you that split is hiding the more important half.

Look at the range, not the single number. If a number rests on fewer than about a hundred cases, treat any difference under ten points as undecided. This applies to our numbers as much as anyone’s, and it is why our report shows a range wherever the cases are few, and says so on the page rather than in a footnote at the bottom.

The one place this gets genuinely misused

These numbers are embarrassingly easy to inflate: you just pick the questions you already win. Ask a hundred questions shaped like “what’s the best tool for [the exact thing you do]” and the result will be magnificent and worthless. The questions have to be the ones a buyer actually types, including the broad ones where you lose, and including comparisons with competitors you’d rather not name.

If a vendor won’t show you the question list, there is no way to check the score. Ours is published with every report, verbatim, including the questions where you place badly.