Growth · Measurement and evidence · 02
Prompt set design
Prompt set design builds the defined set of buyer questions that AI visibility is measured against: the questions real buyers ask at each stage of a decision, phrased the way they ask them, organized into clusters by intent and service, versioned so that measurements over time compare like with like, and reviewed as buyer language changes. It is needed before share-of-voice tracking can begin, when measurement has been based on a handful of guessed prompts, when sales, support and search data show buyers asking questions nobody has written down, or when a content program needs to know which questions it is trying to win. The design produces the documented prompt set with sources and clusters, the engine and schedule plan, and the review cadence.
Scope
What the review covers.
- Question discovery from sales calls, support, search queries, forums and existing prompt data
- Phrasing in buyers’ language, by decision stage
- Clustering by intent, service and audience
- Engine and schedule selection for measurement
- Versioning so changes are recorded
- Review cadence as buyer language shifts
Deliverables
What you receive.
| Deliverable | Form |
|---|---|
| Prompt set | Documented, sourced, clustered, versioned |
| Measurement plan | Engines, schedule, method |
| Coverage map | Prompts against the pages intended to win them |
| Review record | Changes to the set over time |
Prerequisites
What we need from you.
- Access to sales and support knowledge and to search query data
- The services and audiences the program targets
- Agreement on engines and cadence
Related
Where this sits.
- Measurement and evidence Service
- Growth Hub
- Attribution and call tracking Also
- Dashboard Also
By David Selakovic · Published 2026-10-16 · Revised 2026-10-16 · Reviewed every 180 days
Questions
Asked before engaging.
How many prompts should there be?
Enough to cover the intents that matter, organized so results are readable. The number is set per engagement in the proposal.
Why not use generic industry prompts?
Because buyers ask specific questions in specific words. Generic prompts measure a market, not your buyers.
Does the set change?
Yes, deliberately and with a record. Buyer language changes and the set follows, with versioning so trends remain comparable.