Skip to main content

​
Direct answer

Use Avg Position only for SELF mentions with a recorded rank. A lower numeric rank represents an earlier detected position, but the average excludes answers where no eligible SELF position exists. Always pair it with absence and mention metrics.

​
Data required

  • Completed answers with detected SELF mentions
  • Recorded SELF positions greater than zero
  • Eligible-position count and distribution
  • Answers with no SELF mention
  • A fixed provider, model, prompt, country, language, and period scope

​
Workflow

  1. Set a consistent scope in Cockpit > Overview or Cockpit > Compare.
  2. Record Avg Position together with the number of SELF mentions that have a recorded rank.
  3. In Monitor > Prompt Tracking, locate prompts with earlier, later, or missing positions.
  4. Read those answers in Monitor > Response Analysis to confirm that rank is meaningful for the response format.
  5. Separate list-style answers from narrative answers when their position semantics differ.
  6. Compare providers or periods only when the eligible answer populations are comparable.

​
Interpretation

Avg Position is conditional on a recorded SELF mention position. It is not prompt Coverage and it does not include brand absences as a lower rank. An average can therefore improve while your brand disappears from more answers. Use the distribution and underlying answers to distinguish a consistent pattern from a few extreme ranks. There is no universal good position across response formats or categories.

​
Possible next actions

  • Test whether a prompt-specific content gap explains late or missing placement.
  • Compare list and narrative prompts separately.
  • Investigate whether one provider or model accounts for the position shift.
  • Pair a position experiment with Coverage and Mention Rate so absence remains visible.
Treat any expected ranking change as a hypothesis.

​
Limitations

  • Position is not equally meaningful in every answer format.
  • Averages hide dispersion and exclude ineligible observations.
  • Detection can be affected by aliases, tables, and repeated mentions.
  • Position does not measure sentiment, citation, or commercial impact.