AI visibility tools: what they actually measure
Most AI visibility tools measure one thing: how often you are cited. What that number includes, the three things it structurally cannot see, and how to judge one before you pay.
Almost every AI visibility tool measures the same thing: how often your brand is cited when engines answer a set of questions. The differences are in how many engines, how many questions, and whether it tells you what to do about the answer.
What do they all have in common?
A prompt set and a citation count. The tool asks a list of questions across some number of engines on some schedule, records who was named, and charts it.
That is genuinely useful and it is also now available free. Microsoft Clarity added a citation dashboard in May 2026 with no traffic limits, which puts a floor under the whole category.
What can a citation count not tell you?
Three things, and they are the three that decide what you do on Monday. It cannot tell you why you were not cited, which page to change, or whether a change you made worked.
It also cannot see behaviour. The people who do arrive from an answer are invisible to a tool with no tag on your site, so the most engaged traffic you have goes unmeasured.
How should I judge one before paying?
Ask three questions. Where do the prompts come from, because invented prompts measure a market that does not exist. Are citations counted apart from mentions, because a mention with no link earns nothing. And does it publish its method.
Then ask the one most vendors dislike: what does it tell me to change? A number with no instruction is a report, not a tool.
What should it cost?
Published prices in this category run from about $29 a month for a single-engine tracker to $399 for the enterprise-facing tiers, with a free floor underneath from Clarity and Amplitude.
Anything priced above that band should be replacing work you currently pay a person to do, not just charting a number.
Questions
Is a free AI visibility tool good enough?
For a citation count, often yes. Clarity gives it away with no traffic cap. The gap is instruction: it does not score your page, rank a fix, or record whether the fix moved anything.
How many engines do I need?
At least three, because they disagree. Tracking only ChatGPT is the most common mistake and produces a number that swings on one vendor's retrieval changes.
Can I build this myself?
Yes, and people do. It is a prompt set, an API call per engine, and somewhere to store answers. The work that is not trivial is sourcing honest prompts and keeping the run stable enough to trend.
Find out where you stand on your own domain: the free AI visibility check asks all four engines about you, and the content scorer scores any page against the published rubric. Neither needs an account. The full background is what answer engine optimization is.