Skip to main content

​
Direct answer

Compare providers only after matching the prompt set, period, country, language, and answer coverage. Use provider-level Mention Rate and raw counts in Cockpit > Overview, then inspect model-level and answer-level differences.

​
Data required

  • The same answered prompts across the providers being compared
  • Provider and model identifiers
  • SELF and DIRECT brand classifications
  • Mention Rate numerator and denominator by provider
  • Prompt-level Coverage and raw answer counts

​
Workflow

  1. In Cockpit > Overview, fix the period, topic, tag, funnel, country, and language scope.
  2. Compare provider rows using raw answer counts, Mention Rate, and Coverage.
  3. In Cockpit > GEO Matrix, identify the prompts or topics where differences occur.
  4. Separate models within a provider when multiple models are present.
  5. Read representative answers in Monitor > Response Analysis to check mention relevance and wording.
  6. Use Cockpit > Compare with the same filters to add DIRECT competitor context.

​
Interpretation

Product Mention Rate is response-level: answers with a SELF mention divided by answers with at least one SELF or DIRECT brand mention. Product Coverage is prompt-level: distinct answered prompts with at least one SELF mention divided by distinct answered prompts. A provider can lead on one metric and not the other. A provider with incomplete answers or a different prompt mix is not directly comparable. The result describes the selected stored dataset, not the provider as a whole.

​
Possible next actions

  • Test whether a provider-specific topic gap persists with a matched prompt cohort.
  • Inspect the sources and wording behind a low provider result before changing content.
  • Add monitoring for a relevant model when the current sample is incomplete.
  • Revisit DIRECT competitor classification when the Mention Rate denominator looks unexpected.
Treat provider differences as observations to investigate, not proof of a provider’s permanent behavior.

​
Limitations

  • Provider and model output can vary between runs.
  • Availability and source-link behavior can differ by provider, model, and market.
  • Unequal answer volumes distort rankings.
  • Mentions do not indicate endorsement, accuracy, citation, or referral traffic.