GEO overview and metrics
Measure how a brand appears across AI answers and compare the sources shaping those answers.
The GEO overview summarizes how the selected brand appears inside a tracked population of AI answers. Visibility, share of voice, sentiment, position, mentions, and source retrieval each describe a different part of that population. None of them represent the entire public behavior of an AI model.
Read every trend with its prompts, models, topics, tags, markets, competitors, and date range. When a metric changes, inspect the underlying responses and citations before assigning a cause. The overview is most useful for locating a pattern that deserves evidence review, not for replacing that review.
A useful GEO review combines rate, volume, and verbatim evidence. A visibility percentage with only a few answers is fragile, while a large answer set can hide important differences between models or topics. Segment the population when the aggregate changes, then read representative positive, negative, and missing-brand responses to understand the actual language and sources behind the movement.
| Metric | Meaning | Read with |
|---|---|---|
| Visibility | The share of tracked answers in which the brand is mentioned. | Prompt coverage, model mix, and time range |
| Share of voice | The brand's share of mentions within the tracked competitive set. | Competitor configuration |
| Sentiment | The measured tone of brand mentions in stored answers. | Answer text and model |
| Position | Where the brand appears in ranked or ordered recommendations. | Answer structure and sample size |
| Source retrieval | How often a domain or URL is retrieved or cited. | Citation rate and answer count |
Overview layout
The GEO overview includes visibility trend, top brand rankings, source and domain-type distributions, and recent AI answers.
Move from the trend to the ranking and distribution views, then open recent answers that represent the pattern. Source-type and domain summaries help explain where influence concentrates, while answer detail confirms whether the brand language and recommendation context are actually favorable.
Metrics describe the tracked set
Changing prompts, models, topics, tags, countries, or time range changes the population. Record those filters when comparing periods or sharing a report.
When filters or configuration change, annotate the comparison and consider starting a new baseline. A movement caused by a different population is not evidence that public model behavior changed.