# Measure GEO

Track visibility and citations without losing sight of qualified demand, conversion, and revenue.

## Use a metric ladder

| Layer | Example metrics |
| --- | --- |
| Presence | mention rate, recommendation rate, citation rate |
| Positioning | message accuracy, use-case fit, sentiment, competitor share |
| Engagement | AI referral sessions, engaged visits, return visits |
| Conversion | sign-ups, qualified leads, assisted conversions |
| Business | opportunities, pipeline, revenue, sales-cycle influence |

## Separate observation from inference

Direct AI referral traffic is only part of the effect. Buyers may see a recommendation and later arrive through search, direct, or branded channels. Combine referral data with self-reported attribution, CRM notes, branded demand, conversion paths, and customer interviews.

## Create a testing cadence

Use a stable core prompt set for trend comparison and a rotating discovery set for new behaviors. Record the engine, date, locale, account state, and method. Compare clusters and repeated samples rather than celebrating individual answer changes.

## Report what changed

A useful report explains the visibility movement, which sources or claims appeared, what work shipped beforehand, how qualified demand changed, and what the team will test next.