Skip to main content

​
Direct answer

There is no single visibility measure for every decision. Start in Cockpit > Overview, then combine Mention Rate, Coverage, Share of Voice, Citation Rate, and answer context. Keep each metric’s unit and denominator explicit.

​
Data required

  • Completed answers for a fixed period and filter scope
  • SELF and DIRECT brand and source classifications
  • Answered-prompt counts
  • Mention-occurrence counts
  • Provider, model, country, language, topic, tag, and funnel filters

​
Workflow

  1. Set the analysis scope in Cockpit > Overview.
  2. Record every headline metric with its numerator and denominator.
  3. Use Cockpit > GEO Matrix to locate topic and provider differences.
  4. Use Monitor > Prompt Tracking to identify covered and uncovered prompts.
  5. Read representative answers in Monitor > Response Analysis to evaluate context, accuracy, and prominence.
  6. Repeat only with comparable scopes when benchmarking another period or segment.

​
Interpretation

  • Mention Rate: answers with a SELF mention divided by answers with at least one SELF or DIRECT brand mention.
  • Citation Rate: answers citing a SELF source divided by answers citing at least one SELF or DIRECT source.
  • Share of Voice: SELF mention occurrences divided by SELF plus DIRECT mention occurrences.
  • Coverage: distinct answered prompts with at least one SELF mention divided by distinct answered prompts.
Mention Rate and Citation Rate are response-level, Coverage is prompt-level, and Share of Voice is occurrence-level. Do not average or combine them into a new score without an approved methodology.

​
Possible next actions

  • Test whether a topic with low Coverage needs a better prompt set before proposing content.
  • Investigate provider-specific low Mention Rate in the underlying answers.
  • Test a source-page improvement when SELF citations are absent for a recurring factual question.
  • Review DIRECT competitor classification before acting on Share of Voice.
Treat each expected outcome as a hypothesis and preserve the comparison scope.

​
Limitations

  • The dataset covers configured prompts and completed answers, not every AI interaction.
  • Presence does not imply endorsement, accuracy, or purchase intent.
  • Missing or uneven samples can distort cross-provider comparisons.
  • Provider output and citations can vary between runs.