Start with the questions
A hotel might appear when someone asks for a boutique stay in Tokyo and disappear when the question mentions children or a business district. One overall score can hide that difference.
Build your question set around decisions your buyers actually make. Record the topic, location, language, and intent. Keep the questions stable when you compare results over time, and explain when you add or remove any.
Separate presence from preference
A brand can be listed without being recommended. Count those outcomes separately. For our example report, a mention means the brand appears in the answer; a recommendation means the answer explicitly suggests choosing it. Apply the same rules to every response.
Repeat questions because responses can vary. If you ask 20 questions five times each, you have 100 responses, not 100 different questions. A score of 40 means the brand appeared in 40 of those responses. It does not mean the brand owns 40% of the market.
Show the evidence behind the number
A useful report includes the collection dates, platform, question set, scoring rules, and response evidence. Break it down by topic and keep cited URLs available for review. Check whether the sources support the answer before treating a citation as an opportunity.
Use the result to choose an investigation: missing service details, unclear positioning, or a gap in credible coverage. Track referrals and enquiries separately where you can. An improvement in mentions can be worth knowing about without pretending it proves a revenue increase.