Discovery plan — explainable prioritization

Decision to inform

Before extending this fictional prototype, test whether a B2B SaaS PM preparing a planning discussion finds preserved evidence and editable priority inputs more useful than their normal spreadsheet or discovery tool. This is a research plan, not completed research; Northstar accounts and feedback remain fictional.

Assumptions to challenge

• PMs need the original quote and account context at prioritization time.
• Showing impact, confidence, effort, and the score helps a reviewer challenge a decision.
• Explicit tag review is worth the extra step when a suggestion is simulated.
• A group with unique accounts is easier to reason about than duplicated comments.

Participants and session

Recruit five people who currently triage B2B SaaS feedback at least monthly. Include PMs and product-operations practitioners where possible. Use consented 30-minute remote sessions and a sanitized fictional workspace; do not collect customer data, account names, or recordings without separate consent. Give each participant the same scenario twice: once in their normal spreadsheet or discovery method and once in Feedback Nexus. Counterbalance the order across participants, keep the fictional evidence and time box consistent, and ask them to compare related feedback, explain a decision, correct or leave a suggested tag, and decide whether to move a signal to Planned. Use neutral prompts: “What are you looking at now?”, “What would you need to decide?”, and “What, if anything, is unclear?” Avoid telling participants that traceability, grouping, or the score should help.

Signals and falsification

Record completion, elapsed time, explanation quality against the proposed Validation rubric, confusion about simulated tags, requests for missing information, and each participant’s comparison of the two methods. Keep assisted and unassisted completion separate. Evidence against the direction includes participants preferring their existing workflow, treating the score as an automatic answer, being unable to identify source evidence, or finding duplicate grouping less clear. If two or more participants show the same failure, pause feature expansion and revise the workflow before considering import, collaboration, or automated tagging.