Define what you will measure

Choose a few services, locations and questions tied to real client needs. Record which providers, dates and competitor businesses are in scope.

Separate answer evidence from visits

Count mentions, recommendations and citations separately. Site visits and inquiries answer different questions and may be affected by factors beyond search visibility.

Framework exhibit · Not measured data
Three evidence layers, three different questions
Evidence layerWhat it supportsWhat it does not prove
Answer benchmarkPresence in the defined test panelPopulation-wide preference
Platform reportingRecorded website exposureAn explicit business recommendation
Referral analyticsIdentifiable recorded visitsAll AI-influenced demand

Original editorial framework. This exhibit contains no observed client results, performance statistics, or market estimates.

Review the comparison consistently

Repeat the same questions where possible. Retain answer records and flag changes in the provider, phrasing or location before comparing results.

Decide what counts before testing

Define a mention as a recognizable business name, a recommendation as explicit advice to consider the business, and a citation as a link used to support an answer. Count each separately and flag ambiguous names. Avoid comparing headline percentages if the underlying questions changed.

Sources & references

Official documentation reviewed November 13, 2026. Platform features and policies can change. The linked sources support the cited facts; proposed workflows are editorial analysis.

  1. Google Search Central: AI features and your websitehttps://developers.google.com/search/docs/appearance/ai-features
  2. OpenAI: Overview of OpenAI Crawlershttps://developers.openai.com/api/docs/bots