Organise the businesses you measure
Use the brand selector to choose the business whose evidence you want to inspect. My websites gives you the portfolio context and available website analysis. Confirm the selected domain before reading a chart or creating a report. A similar brand name or an old domain can produce a misleading interpretation even when the underlying response is accurate.
For agency work, keep each client’s business identity and question set clear. Name saved views so another colleague can recognise the client, market and reporting period. The Agency subscription is a volume option, not a requirement to serve clients; the feature set is the same on every tier.
Build questions from real customer decisions
- List the customer tasks you already support: discovery, comparison, purchase, implementation and ongoing use.
- Choose a small set of questions that would change a business decision. Include category questions, alternative comparisons and questions about evidence or suitability.
- Save the exact wording in All prompts. Use topics and tags to group questions by product, audience or intent.
- Record market, language and source scope. A translated question or changed provider set starts a different comparison.
- Review Suggested and Archived views where available. Saving or organising a prompt does not itself establish a measured result.
Use Discovery as a research queue
Discovery and Fanouts can expose returned research queries and supplied keyword candidates. These help you find questions to investigate. A query returned during one provider run is not a count of how often customers ask it. Topic-demand estimates are research signals with a stated method, not a measurement of private assistant conversations.
Review candidates for relevance, duplication and commercial usefulness. Promote a candidate into your fixed set only when you can explain who asks it and what decision it supports. Keep the original version of a question when changing its wording, so earlier results are not silently compared with a different test.
Budget repeat use before enabling schedules
Calculate searches as questions × market/language combinations × runs during the billing period. Selected available sources are included within the quoted complete-search unit. For example, 20 questions in one market repeated four times consume 80 complete searches. That leaves 20 of a Professional plan’s 100 monthly credits for additional investigation. This is a planning example, not an automatic schedule or a guarantee of results.
Five clients with 15 questions each, measured four times, require 300 complete searches in one market. A month with five scheduled runs would require 375. Check actual billing-period dates and calendar frequency; “weekly” does not always mean four runs. Keep an allowance for reruns or new questions.
Maintain the set without losing history
Pause tracking for questions that no longer support a decision. Archive records you want out of the active library while retaining their history; use restore when you need them again. Treat Trash and permanent removal separately. Review the schedule, scope and available credits before enabling new collection. Scheduled delivery of an existing report is different from a schedule that collects new answers and consumes credits.
Your next step
Create a question register with an owner, purpose, market, frequency and monthly search total.
Continue to the next guide → · All Academy lessons