Skip to main content

​
Direct answer

Filter to prompts assigned to the purchase-intent or BOFU stage. Measure SELF presence, then read the answers to determine whether the brand is recommended, compared, excluded, or merely named.

​
Data required

  • Answered prompts consistently assigned to the purchase-intent or BOFU stage
  • SELF and DIRECT competitor classifications
  • A fixed provider, model, country, language, topic, tag, and period scope
  • Coverage, Mention Rate, Share of Voice, position, and raw eligible counts
  • Full answer text and exposed citations

​
Workflow

  1. In Workspace > Prompts, confirm that purchase-intent prompts ask for a concrete decision without forcing your brand into the wording.
  2. In Cockpit > GEO Matrix, apply the purchase-intent or BOFU filter.
  3. Record Coverage and Mention Rate with their prompt-level and response-level denominators.
  4. In Cockpit > Compare, inspect DIRECT competitors and Share of Voice under the same scope.
  5. Read missing and present cases in Monitor > Response Analysis to verify recommendation context.
  6. Review exposed sources in Monitor > Source Explorer when an answer relies on outdated or unsupported facts.

​
Interpretation

Presence is not purchase intent, endorsement, or conversion. Product Coverage shows the share of distinct answered prompts with at least one SELF mention. Product Mention Rate shows SELF-mentioned answers among answers with a SELF or DIRECT mention. Avg Position, when available, is conditional on recorded SELF positions and excludes absences. Pair it with Coverage so a better average cannot hide disappearing mentions.

​
Possible next actions

  • Test whether a product or pricing page clearly supports a recurring selection criterion.
  • Correct factual gaps found in decision-stage answers.
  • Investigate provider-specific absence before changing broad positioning.
  • Track the same purchase-intent prompt cohort after a page update.
Treat expected visibility or recommendation changes as hypotheses.

​
Limitations

  • Funnel-stage assignments are editorial, not observed user intent.
  • A monitored prompt is not evidence of real purchase demand.
  • AI recommendation does not prove traffic or conversion.
  • Providers can change answers and citations between runs.