Classify AI visibility evidence before calculating anything.

Use this framework without an email. It does not query answer engines from this page and therefore reports no current brand result. It helps you design a reproducible probe set and honest evidence labels.

Use the Framework Read the GEO Guide

Nothing publishes without approvalPaid product accessNo ranking guarantees

Three non-interchangeable classes

Synthetic probe · observed citation · real referral

Classification worksheet

Label each record before aggregation.

Synthetic probe

Engine/model and access mode
Exact prompt and cohort
Location, account state and timestamp
Captured response and repetition count

Observed citation

Quoted answer passage
Cited brand and destination URL
Whether the citation supports the claim
Screenshot or response artifact

Real analytics

Authorized provider and property
Referral source and landing page
Sessions or conversions
Attribution and consent limitations

Example result

Report a sample, not a universal visibility score.

Example: ‘In 30 approved prompts run twice on one engine on 20 July, 8 responses mentioned the brand and 3 cited an owned URL.’ Do not rewrite that as ‘27% AI market visibility.’

Limitations

Responses vary by model, account, region, personalization and time. Probe coverage reflects the designed sample. Referral analytics may undercount or misclassify answer-engine journeys.

Paid deployment · human authority

Operate a timestamped methodology instead of a vanity score.

Choose paid AI-visibility capacity when your prompt cohorts and evidence policy are approved.

Choose Your Plan Book a Demo