Feedback Nexus — decision brief

Compare customer feedback while keeping the original quotes and affected accounts visible. Record why one problem deserves attention before choosing what to work on.

The product choice: Use scores to support a discussion while keeping the source evidence open to challenge. Try the sample · Follow the walkthrough.

Experience and decision

My work on product requirements, customer requests and roadmap priorities has required separating a requested feature from the underlying problem. This is a separate fictional portfolio exercise; its records, rules and results do not reproduce an employer system.

Decision to challenge: When should a lower-scoring problem take priority? The reviewer needs to weigh the source evidence, confidence and strategic context rather than let the score make the roadmap decision.

User and problem

A B2B SaaS PM preparing a planning discussion needs to compare customer feedback without losing the original words or account context. In the fictional Northstar workspace, import recovery and saved reporting compete for attention. The product question is whether a reviewer can inspect the evidence behind a planning recommendation rather than inherit an unexplained rank.

Product choice

The prototype is an evidence board, not an automatic roadmap. It preserves individual signals, links related feedback, shows unique affected accounts, exposes impact, confidence, and effort, and lets the PM record a rationale and planning state. The visible comparison is affected accounts × impact × confidence ÷ effort; it starts a conversation and does not make a decision.

F01 and F02 demonstrate the choice. Their original quotes remain separate while the group shows the four-account union: Alder Labs, Finch Studio, Cedar Systems, and Pineworks. F04 remains a distinct saved-report alternative. Simulated tags are editable support, not evaluated assistance, and their review is optional.

Cost and boundary

Traceability costs time: the PM must inspect evidence and explain a choice. Unique account reach also omits contract value and strategic importance. Browser-local state, export, Undo, and reset make a safe portfolio walkthrough possible, but do not provide shared history, permissions, or operational durability. The fictional sample does not connect to customers, a CRM, or a model.

Evidence and next investment

Software checks show that the documented workflow and recovery paths ran; they do not show demand, usability, or outcomes. No customer interviews or human study have occurred. The next investment is a five-participant comparison with each person’s current method, using consented fictional data and the published scoring protocol. Expand scope only if participants can trace the source evidence, explain the score and simulated-tag boundary, and make a justified decision without critical misunderstanding. The walkthrough, PRD, decisions, and validation plan expose that testable product case.

Before adding ingestion or automatic classification, test whether source-preserving review changes the quality of a planning decision compared with the current spreadsheet or backlog. Ask why import recovery should outrank saved reporting and what evidence could reverse that choice. Account reach omits revenue, urgency and support burden, so a higher score alone cannot justify investment. If reviewers still follow the rank without challenging confidence or effort, simplify the scoring emphasis. Commercial discovery must separately establish whether the review benefit exceeds the work of maintaining the evidence board.

My role as Product Manager

I defined the problem, prioritization and scope, workflows, fictional sample data, acceptance criteria, and evaluation plan. AI tools assisted with implementation and verification; I do not claim manual authorship of the application code.