Skip to main content

​
Direct answer

Choose one DIRECT competitor and compare answers or prompts where both brands appear, only your brand appears, only the competitor appears, or neither appears. State the unit before calculating overlap, then read the answers to determine whether co-appearance is a genuine recommendation.

​
Data required

  • Completed answers for a fixed prompt, provider, model, market, and period scope
  • Correct SELF and selected DIRECT competitor classifications
  • Mention detections for both brands
  • A chosen unit: answers or distinct answered prompts
  • Full answer text and exposed citations

​
Workflow

  1. Confirm that the selected brand is classified as a DIRECT competitor.
  2. In Cockpit > Compare, apply the competitor and analysis filters.
  3. Export or inspect the underlying records and assign each unit to both, SELF-only, competitor-only, or neither.
  4. In Cockpit > GEO Matrix, locate the topics and prompts where the pattern is concentrated.
  5. Read the matching answers in Monitor > Response Analysis to distinguish recommendation, comparison, criticism, and incidental mention.
  6. Repeat for another period only with the same unit, prompt cohort, and competitor classification.

​
Interpretation

Answer-level overlap and prompt-level overlap are not interchangeable. A prompt can have separate answers where each brand appears without any single answer containing both. Product Share of Voice counts SELF and DIRECT mention occurrences; it does not measure co-appearance. Product Mention Rate is response-level and also does not provide an overlap cohort by itself.

​
Possible next actions

  • Test content that addresses a criterion recurring in competitor-only answers.
  • Investigate a provider or topic where co-appearance differs from the rest of the dataset.
  • Correct competitor aliases or relationships when detections are incomplete.
  • Track the same overlap cohorts after a positioning experiment.
These actions test hypotheses and do not guarantee recommendation changes.

​
Limitations

  • Co-appearance does not mean the brands are equally recommended.
  • Mention detection can be affected by aliases and ambiguous names.
  • Small cohorts can change sharply between runs.
  • Qwairy observations do not represent every provider interaction.