How it works

Three tiers of measurement. Seven layers of scoring. One honest report.

We never blend numbers across AI engines or inflate visibility with questions that already include your brand name. Every figure in the report carries the tier that produced it and the layer it belongs to.

Tier 1 · The AI models, asked directly

How the four AI providers describe your brand

We ask Claude, ChatGPT, Gemini, and Perplexity the same 100 customer questions your buyers type. In your market, in your language, framed like a real buyer. We capture every answer verbatim.

  • Same questions, same day, across all four AIs. Apples to apples.
  • Each AI is called with the country, region, and language of your audience.
  • We capture every citation so you see which sources the AI trusts.
Tier 2 · Google AI Overviews

What the AI answer at the top of Google search says

For every question, we check whether Google's AI Overview shows up. If it does, we capture what it says. If it doesn't, we say so. Silence is not the same as being ignored.

  • Run from a Netherlands IP with Dutch and English locale by default.
  • AI Overview trigger rate reported separately from mention rate. 'No overview' is not 'not mentioned'.
  • Cited sources parsed and grouped so you see which domains keep appearing.
Tier 3 · ChatGPT as customers use it

What a real customer sees when they open ChatGPT

This tier separates a real audit from a generic tool. We run every question through a logged-in ChatGPT session from a Netherlands IP, paced like a person typing. Every answer ships as a screenshot in your report.

  • The developer API and the consumer app disagree on 76% of brand mentions.
  • Screenshots ship with the report as evidence. You can hand them to a skeptic.
  • This is what your buyer sees. Most tools skip it.
The seven layers

Score every layer separately, never a single blended number.

Generative-search metrics are random variables, not fixed values. Collapsing them into one score multiplies noise and hides the only thing you can act on: which layer is broken.

  1. L1
    Site readiness
    Can AI engines reach, parse, and trust your site?
  2. L2
    Presence
    Do you show up in the answer? Measured when the question doesn't include your name.
  3. L3
    Preference
    Are you recommended, or listed among options?
  4. L4
    Citation
    Is your own domain among the sources the AI cites?
  5. L5
    Accuracy
    When it describes you, does it get the facts right?
  6. L6
    Consensus
    Do independent sources corroborate you and agree with each other?
  7. L7
    Business impact
    Does any of this translate into signups, sales, or retention?

Read the full methodology at /methodology.

See how AI describes your brand today.

Get my free snapshot