Recruit for recent decisions, not opinions
Ask for people who evaluated a category in the last six to twelve months — buyers, influencers, and honest non-buyers. Opinion panels produce polite abstractions. Recent-decision panels produce timestamps, artifacts, and regret. klitflempot recommends screening for role clarity: economic buyer, technical evaluator, end-user champion, or procurement steward. Mixing roles in one call without labeling them muddies the evidence.
Offer a clear timebox (usually 45–60 minutes) and a research purpose that is not a disguised pitch. When klitflempot coaches teams, we insist the recruiter never promise product access as the reason for the call. That incentive warps answers toward flattery.
Open with the last evaluation, not the ideal future
Start with: “Walk me through the last time your team evaluated tools in this space.” Then follow the chronology. Who initiated? What triggered urgency? Which documents circulated? Which meeting killed momentum? Ideal-state questions (“What would a perfect solution look like?”) invite brochure language. Chronology invites memory.
Probe for silent actors. Many B2B deals stall because a named “decision maker” never held veto power. Pair interview insights with a committee map so quotes are attributed to roles, not just names.
Veto probes and workaround language
Useful prompts include: “Who could have stopped this even if you loved it?” “What did you try before buying?” “Which requirement looked firm on paper but softened under deadline?” Workarounds are gold. They reveal the job people actually hire a product to do, which later informs positioning synthesis.
Capture verbatim phrases. Buyers often reuse internal metaphors — “air traffic control for tickets,” “spreadsheet with a conscience.” Those phrases outlive your interview deck. Tag them in notes so the claude cadence can promote them into claim drafts without inventing slogans.
What klitflempot refuses to claim from interviews
Interviews do not prove market share, conversion lift, or ROI. klitflempot’s editorial standard forbids dressing ten conversations as “representative statistics.” Ten conversations can generate sharp hypotheses, message tests, and friction clusters. They cannot authorize a percentage on a homepage hero. For how those clusters form, read insight patterns.