Fluens AI
AI visibility tracking

A rate, not a single answer

Assistants sample their output, so the same question gives different results on different runs. We ask each prompt repeatedly on every engine and show you how often you are actually named.

Request early access
No newsletter. One email when we widen access.
How it measures

Three numbers that behave differently

Named is not the same as recommended, and neither is the same as being described correctly.
01

Named

The share of answers that mention you at all. The blunt measure, and the one that moves first when third-party coverage changes.

02

Recommended

The share that present you as a suggested option rather than listing you in passing. This is the number that tracks revenue most closely.

03

Accurate

Whether what the assistant says about your pricing, features and positioning is currently true. A confident wrong answer costs you the sale as surely as no answer.

Provenance

Every number says where it came from

Most tools in this space present inferred data as measured data. Each answer here carries how it was collected and what the provider did or did not report.
observed queries — the provider reported the searches it ran
queries not exposed — it searched, but does not report what for
no search performed — answered from memory, nothing retrieved
capture failed — we could not read the metadata for this run

Common questions

What is AI visibility tracking?
Measuring how often AI assistants name your brand when someone asks a question your customers would ask. Because assistants sample their answers, it is a rate across repeated runs rather than a single position — asking once and seeing your name proves very little.
How many runs do you need before a number means anything?
More than one, and the returns fall off quickly. We default to several runs per prompt per engine, which is enough to tell a reliable position from a lucky answer. The sample size is shown next to every figure so you can judge it yourself.
Which engines do you track?
ChatGPT, Gemini, Perplexity and Google's AI surfaces. They draw on different sources and disagree with each other more than most people expect, so being named in one is not being named in all.
Is this an AI visibility score?
We show rates, not a composite score. A single blended number hides which question you are losing and which engine you are losing it on, and it cannot be traced back to an answer you can read. Every figure here links to the samples behind it.
Would rather we ran it for you? The audit is free →