The per-provider breakdown

One brand, several assistants

Different AI assistants can talk about your brand differently. The per-provider breakdown shows your results for each assistant your scan queried, for example ChatGPT, Claude and Gemini.

Why it's useful

• An assistant where you score low points to where you're least visible, a place to focus.
• Big differences between providers can hint at where your presence is thin or your story is unclear.
• If your audience leans on one assistant in particular, you can prioritize it.

Do the assistants agree?

The breakdown above tells you how often each assistant names you. It cannot tell you whether they name you for the same questions, and two assistants can land on the same percentage while answering about completely different things.

Engine agreement compares them prompt by prompt, and sorts every prompt into one of three groups:

• Every engine named your brand. Presence you can count on.
• Only some did. Your presence there depends on which assistant someone happens to use.
• None did. A visibility gap, which the mention-rate signal already covers.

The middle group is the one to watch. A brand can hold a healthy overall mention rate that is mostly made up of prompts a single assistant answers with its name. That presence is fragile: each assistant reads a different set of sources, so earning a mention that works for one does little for the others. The fix is breadth, not depth on any single site.

A prompt an assistant never answered is left out of the comparison rather than counted as a miss, so a provider outage never shows up as a brand problem. When that happens the card says how many prompts it set aside.

Two more comparisons sit alongside it:

• Who each assistant sees as your rivals. A competitor every assistant names is your real category competition. One that only a single assistant names says more about that assistant's sources than about your market.
• Whether they read the same web. How much the pages the assistants retrieved overlap. This only exists when the scan ran with web search: without it there are no retrieved sources to compare, and we say so rather than showing a zero.

Choosing what's included

You decide which assistants a scan uses. If a provider you expected is missing, check your selection at the company, project or scan level (see Choosing providers and Provider locks).