What do we do when a model gets our brand wrong?

What this checks

The recovery sequence for a model stating something false about you, in the order that actually changes the answer.

When to ask this

Best once you have some organic traffic and at least one AI-visibility measurement on file — useful across Some Traction and Heavy Traction stages. This is a live diagnosis over your own data, not a static report, so there's nothing to check before you've been measured at least once.

When this can't anchor to your data

This always returns its recovery sequence — nothing about your account makes it withhold the rules. It anchors when a prompt panel exists, is grounded, and has answers on file; otherwise it says so directly, and which of the two is missing changes what to do next:

Panel exists but ungrounded:

The visibility run on file answered from what the models already knew rather than from anything they retrieved, so it carries no sources to trace a false claim back to. A grounded run is what turns this from a procedure into an incident log.

No panel at all:

Nothing here records what AI engines currently say about this brand, so this playbook cannot point at a specific wrong answer. A frozen prompt panel is what turns it from a procedure into an incident log.

How to ask it in chat

"what do we do when a model hallucinates about our brand"

What you'll get back

A direct answer computed from measurements already on file — no extra cost, no new data is purchased to answer this.