Start with silent clustering to avoid anchoring, then invite rounds of renaming and merging. Photograph each stage to preserve the journey. When remote, reproduce the flow in Miro or FigJam, using color rules to mark evidence strength and uncertainty without losing momentum.
Create a small, memorable codebook with definitions, examples, and counterexamples. Distinguish pains, desires, workarounds, and opportunities. If multiple people code, track inter-rater agreement informally to catch drift. Revisit categories weekly, merging or splitting codes as new interviews stretch your understanding.
Edge cases often point to blind spots or new segments. Mark surprising quotes, conflicting behaviors, or mismatched expectations with a distinct tag. Do not smooth them away; pull a thread. Many breakthrough ideas start as a stubborn, uncomfortable note nobody can categorize.
Select a model your team actually understands. Simplicity helps adoption, while explicit confidence prevents false precision. Start with RICE or an impact-effort grid, then iterate. Publish examples of scored items so newcomers learn faster and debates move from gut feelings toward shared criteria.
Invite engineering, design, support, and sales to score a handful of candidates together. Use timeboxes, posters of definitions, and real quotes to anchor the discussion. Disagreement is a feature; capture rationale alongside scores, then summarize trade-offs in a short, shareable decision log.