Skip to main content

​
Direct answer

Answer this at the answer level. Build two cohorts within the same filtered dataset: answers with both a SELF mention and a SELF citation, and answers with a SELF mention but no SELF citation. Do not subtract Product Citation Rate from Product Mention Rate because the two metrics use different denominators.

​
Data required

  • Completed answers for one consistent period and filter scope
  • Detected SELF and DIRECT brand mentions
  • Citation URLs classified as SELF or DIRECT sources
  • The corresponding answer text for context
  • Provider, model, country, language, topic, and tag filters

​
Workflow

  1. Set the intended scope in Monitor > Response Analysis and identify answers with a SELF mention.
  2. Open those answers and record whether each also contains a citation to a SELF source.
  3. Use Monitor > Source Explorer to inspect the SELF domains and URLs cited in the same scope.
  4. Export answer-level data when you need a reproducible cohort count across many answers.
  5. Review the mention and citation context. A linked source may be unrelated to the sentence that mentions your brand.
  6. Repeat with the same filters for a comparison period or provider.

​
Interpretation

Product Mention Rate is answers with a SELF mention divided by answers with at least one SELF or DIRECT brand mention. Product Citation Rate is answers citing a SELF source divided by answers citing at least one SELF or DIRECT source. Both are response-level, but their eligible answer populations differ. For this use case, the most direct unit is therefore a count or proportion among answers with a SELF mention. Keep the numerator and denominator visible. A SELF citation does not imply a positive mention, and a positive mention does not imply a citation.

​
Possible next actions

  • Test whether a clearly sourced page for a recurring claim is associated with more SELF citations in later comparable answers.
  • Review unlinked mentions to identify facts that your site does not currently document with verifiable evidence.
  • Test whether consolidating duplicate or outdated pages gives providers a clearer canonical source.
  • Investigate a provider-specific gap before making a site-wide change.
These are hypotheses. Compare later observations under the same prompt and provider scope.

​
Limitations

  • Qwairy records citations returned with stored answers; it does not expose every source used to generate an answer.
  • Citation and mention extraction can require manual review for aliases, redirects, and ambiguous domains.
  • A co-occurring citation does not prove that the source caused the mention.
  • Providers differ in whether and how they expose source links.