Skip to main content

​
Direct answer

Use topic filters in Cockpit > GEO Matrix to find where your brand is observed, then verify the underlying prompts and answers. Compare both prompt-level Coverage and response-level Mention Rate. They answer different questions.

​
Data required

  • Answered prompts assigned to stable topics
  • SELF and DIRECT competitor classifications
  • A fixed period, provider or model set, country, and language
  • Topic-level raw answer and prompt counts
  • Representative answer text for each topic

​
Workflow

  1. In Workspace > Topics, review topic names and prompts with no topic assignment.
  2. In Cockpit > GEO Matrix, set a consistent period and filter scope, then compare topic rows.
  3. For each topic, record Coverage, Mention Rate, and their raw numerators and denominators.
  4. In Monitor > Prompt Tracking, identify which prompts account for presence or absence within that topic.
  5. Read the matching answers in Monitor > Response Analysis to distinguish relevant mentions from incidental ones.
  6. Use Cockpit > Compare with the same topic filter when competitive context is needed.

​
Interpretation

Product Coverage is distinct answered prompts with at least one SELF mention divided by distinct answered prompts. It is prompt-level. Product Mention Rate is answers with a SELF mention divided by answers with at least one SELF or DIRECT brand mention. It is response-level. A topic can have broad Coverage because your brand appears at least once for many prompts while still having a lower Mention Rate across answers. Conversely, a narrow topic can show a high rate from a small sample. Inspect counts before ranking topics.

​
Possible next actions

  • Test content that answers a missing prompt with specific, verifiable evidence.
  • Split a broad topic when several distinct buyer decisions are being blended.
  • Add a missing category-neutral prompt when the topic is underrepresented in the monitored set.
  • Review competitor classification when a DIRECT brand is missing from the denominator.
Treat the expected effect as a hypothesis and compare a stable cohort after the change.

​
Limitations

  • Topic results depend on your prompt assignments and do not measure the full market.
  • Unequal prompt or answer volumes make raw percentages hard to compare.
  • Presence does not indicate endorsement, accuracy, or commercial impact.
  • Provider output can vary between runs.