Skip to main content

​
Direct answer

Filter to prompts assigned to the consideration or MOFU stage, then inspect both presence metrics and the full answers. A SELF mention does not by itself mean that the provider recommends your brand.

​
Data required

  • Answered prompts consistently assigned to the consideration or MOFU stage
  • SELF and DIRECT competitor classifications
  • A fixed provider, model, country, language, topic, tag, and period scope
  • Mention Rate, Coverage, Share of Voice, and raw components
  • Answer text showing comparison criteria and recommendation context

​
Workflow

  1. In Workspace > Prompts, review the funnel-stage assignments for comparison and evaluation prompts.
  2. In Cockpit > GEO Matrix, filter to the consideration or MOFU stage and record Coverage and Mention Rate.
  3. In Cockpit > Compare, use the same filter to review Share of Voice and DIRECT competitor presence.
  4. In Monitor > Prompt Tracking, locate prompts where your brand is absent or appears inconsistently.
  5. Read those answers in Monitor > Response Analysis and classify the context as included, compared, recommended, discouraged, or incidental.
  6. Compare providers or topics only with matched prompt and answer coverage.

​
Interpretation

Product Coverage is prompt-level, while Product Mention Rate is response-level. Product Share of Voice counts mention occurrences. None of these metrics alone measures recommendation. Use answer context to determine whether your brand is genuinely considered and which criteria are applied. A high aggregate can be driven by repeated mentions in a narrow set of prompts.

​
Possible next actions

  • Test whether a comparison page addresses a recurring criterion with verifiable evidence.
  • Investigate a topic where your brand is mentioned but not shortlisted.
  • Correct outdated claims found in observed answers or exposed sources.
  • Track a fixed consideration-stage cohort after a positioning change.
Treat each action as a hypothesis; providers control future outputs.

​
Limitations

  • Funnel stages reflect your prompt taxonomy and require consistent judgment.
  • Recommendation context cannot be inferred from mention counts alone.
  • Provider and model output can vary between runs.
  • The monitored prompt set does not represent every consideration journey.