Tracking AI visibility for several clients at once

This is the thing that catches an agency first. A visibility figure is a share of answers to a set of questions, so a client on thirty questions about payment software and a client on thirty about running shoes have two figures that share a scale and nothing else. Reporting them side by side as a league table of your roster tells somebody the wrong thing in a slide they will forward.

What IS comparable across clients is the shape rather than the level: which step is weakest, whether a movement is larger than the measurement’s own margin of error, and whether a client is being named but not chosen. Those are the same question asked of every account, so they rank honestly.

An agency does not open ten dashboards on a Monday. The useful surface is one screen per roster that answers where the attention should go, which is a different question from what each score is. A row that says a client’s weakest step is Trusted and that it has moved by less than the margin of error is actionable. A row carrying only a number is a number.

We render that as a portfolio across every brand in the account, with each brand’s five steps, its band and its movement between measurements. It is the one place the roster is read as a roster.

A measurement costs real money to run, so every tool has some allowance behind it. The question to ask is what that allowance is counted against. Counted per account, ten brands sharing two measurements a month means one brand measured twice and nine measured never, and the page selling it says ten brands.

Ours is counted per brand: each brand on an account gets that level’s included measurements, and anything bought on top is a shared pool the whole account can draw on. That is worth asking of anybody, because the answer changes what you are actually buying by an order of magnitude.

The first is above. The second is the one nobody asks until it matters: an agency relationship ends, and the measurements taken during it are the client’s record of what happened. Ask what leaves with them. Our exports are per brand and carry every stored answer word for word, so the evidence travels rather than the summary.

Two things we do not offer, said here rather than discovered later. Reports do not carry your own branding, and there is no API to pull readings into your own reporting. If either is a requirement, it is worth ruling us out on it now.