Prompt Set Design for GEO Research

Start from interviews, search and site data, sales calls, support questions, and category knowledge. Group prompts by task and stage, then vary only dimensions that reflect real audiences, constraints, or markets. Brainstormed prompts are hypotheses until customer evidence supports them.

Include branded, unbranded category, problem, use-case, comparison, risk, and implementation questions. Tag each prompt by persona, stage, geography, product, and value. Keep a stable core for trend analysis and a rotating exploratory set for emerging language.

Do not overrepresent prompts where the brand already performs well. Record exclusions and known blind spots, particularly for languages or customer groups the research team cannot evaluate reliably.

Assign prompt IDs, preserve wording, document additions and retirements, and separate genuine trend changes from panel changes. Revalidate the set when products, competitors, market language, or platform behavior shifts materially.

Q: How many prompts are enough? A: There is no universal number; use enough to represent priority decisions while retaining a cadence the team can review consistently.

Q: Should prompts name competitors? A: Include realistic named comparisons, but keep them distinct from unprompted category discovery.