Measurement

AI visibility tools measure. They don't move revenue.

By Arnav Mukherjee, founder of TofuBofu · July 24, 2026

Here is a finding the AI visibility category does not put on its landing pages. Reviewers testing these tools reported that more AI brand mentions did not reliably produce more traffic or conversions, and that the monitors did not provide any tools to actually improve visibility. The measurement was accurate. The outcome it was supposed to predict just did not follow.

I build a tool in this space, and I want to sit with that rather than skate past it, because it is the most important thing to understand before you pay for anything here. Almost every tool in this category sells you a number. Very few of them do the thing that would actually change the number, or your revenue. Those are not the same product, and the gap between them is where most of the money gets wasted.

The category sells measurement

Open any AI visibility tool and the promise is a dashboard: here is your share of voice, here is where ChatGPT names a competitor instead of you, here is a trend line. That is genuinely useful. You cannot fix what you cannot see. But notice what the product actually is. It is a mirror. It reflects your position back at you in high resolution, and then it stops. The reviewers who found no link between mentions and revenue were not testing broken tools. They were testing tools working exactly as designed, which is to say, tools that measure.

Why the number and the revenue come apart

A visibility score can rise on mentions that never touch a purchase. You get named in a low-intent, informational answer, the number ticks up, and no buyer was anywhere near a decision. Meanwhile the answer that actually matters, the one a buyer sees when they ask which vendor to choose, still names someone else. A blended score does not distinguish those two mentions, so it can look healthier while the commercially important gap sits untouched.

And even the right mention does not always send a visitor, because AI answers often resolve the question in place. So the chain from mention to click to revenue is leaky at every link. This is why watching the number is not the same as moving the outcome, and why a tool that only watches can leave you diligently tracking a metric that is not doing anything for you.

What actually moves the needle

The lever is not the dashboard. It is getting named in the specific buying-intent answers where your buyers decide, and building enough corroboration across the sources AI reads that the engines are confident enough to name you there. That means producing the content and schema that answer those exact questions, spreading your presence across several sources rather than one, and then re-checking whether the answer actually changed. Profound's research found that brands present across four or more platforms are about 2.8 times more likely to be cited. The needle moves when you act on the gap, not when you admire it.

This is the half most tools leave to you. For a company with a marketing team, that can be fine, the team does the acting. For a founder or lean team without one, a measure-only tool hands you a precise description of a problem you still cannot solve. The measurement was never the hard part. The fixing is.

And be honest about attribution

One more thing, because the opposite mistake is to over-promise. Nobody has clean revenue attribution from AI visibility yet. AI referral traffic is systematically undercounted, and the models to tie a mention to a dollar do not really exist at this stage. So be skeptical of any tool, including this category's louder marketing, that implies a tidy revenue figure. The honest approach is to act on the gaps that matter and track leading indicators you can actually verify: does AI name you in the buying questions your customers ask, and does that change after you do the work. That is closer to a revenue lever than a mention count, without pretending the attribution problem is solved.