Skip to main content

​
Direct answer

Prioritize providers according to your audience and decision context. Combine Qwairy visibility data with connected referral analytics and competitive observations. Do not infer priority from provider popularity alone, and keep the distinction between a provider family and a specific model.

​
Data required

  • Completed answers for the same prompts across the providers or models being compared
  • Provider-level metrics and raw answer counts from Cockpit > Overview
  • Referral sessions by source from Measure > Referrer Analytics, when an analytics integration is connected
  • Your target countries, languages, audiences, and buyer journeys
  • Current provider and model settings from Workspace > Monitoring

​
Workflow

  1. In Cockpit > Overview, select a fixed period, prompt set, country, and language.
  2. Compare providers using both raw answer counts and product metrics. Exclude providers with missing or materially smaller samples from a direct ranking.
  3. In Cockpit > Compare, inspect whether DIRECT competitors appear differently under the same provider and model filters.
  4. If Referrer Analytics is configured, compare observed AI referral sessions with visibility. Keep traffic and visibility as separate measures.
  5. Read representative answers in Monitor > Response Analysis to understand what each aggregate hides.
  6. In Workspace > Monitoring, adjust provider or model coverage only after documenting the evidence and expected trade-off.

​
Interpretation

Product Mention Rate is response-level: answers with a SELF mention divided by answers with at least one SELF or DIRECT brand mention. Product Coverage is prompt-level: distinct answered prompts with a SELF mention divided by distinct answered prompts. A provider can have high Coverage but a lower Mention Rate, or send observed referral traffic despite limited monitored visibility. These are different populations. Referral sessions show visits attributed by the connected analytics source; they do not measure every user interaction with a provider.

​
Possible next actions

  • Test broader monitoring on a provider that is audience-relevant but has an incomplete answer sample.
  • Test whether a low-visibility, observed-traffic provider deserves more prompt coverage.
  • Reduce frequency for a low-evidence provider as a reversible credit-allocation experiment.
  • Separate models within one provider when model-level behavior is masking meaningful differences.
Treat prioritization as a hypothesis and review it after collecting a comparable dataset.

​
Limitations

  • Provider and model availability can depend on the current product configuration and market.
  • Referrer Analytics requires connected analytics data and only covers attributable visits.
  • Unequal prompt, model, country, or time coverage invalidates a simple percentage ranking.
  • Visibility, citations, and referral traffic are associated observations; none proves that one caused another.