How to Handle Feedback and Iteration with pm-product-discovery Skills

The pm-product-discovery module provides a command-driven workflow that transforms raw customer feedback into prioritized features through a five-step loop involving interview summarization, assumption validation, and opportunity mapping.

The phuryn/pm-skills repository offers a structured approach to product discovery through modular skill files and executable commands. By leveraging the pm-product-discovery skills, product managers can systematically process interview transcripts, survey results, and support tickets into validated assumptions and ranked backlog items without manual documentation overhead.

Understanding the pm-product-discovery Architecture

The repository separates concerns into two distinct layers: skills and commands. Skill files are markdown documents located in pm-product-discovery/skills/ that define reusable, step-by-step workflows for specific product management tasks. Command files in pm-product-discovery/commands/ serve as the user-facing interface, handling parameter parsing, file uploads, and workspace markdown generation while invoking the underlying skills.

This architecture ensures that complex logic—such as JTBD extraction or ICE scoring—remains centralized in skill definitions, while the commands provide a simple CLI interface for daily use.

The 5-Step Feedback Iteration Loop

When new customer feedback arrives, run it through this predictable workflow to maintain alignment with real user needs.

Step 1: Ingest Raw Feedback with summarize-interview

The summarize-interview skill reads unstructured transcripts and produces structured summaries. According to pm-product-discovery/skills/summarize-interview/SKILL.md, the output extracts Jobs-to-be-Done (JTBD), pain points, satisfaction signals, and immediate action items.

Use the /interview summarize command to process transcripts:

/interview summarize [paste your interview transcript here]

The command generates a markdown summary following the template defined in the skill file, creating a structured foundation for downstream analysis.

Step 2: Surface and Validate Assumptions

Raw feedback contains hidden assumptions that carry implementation risk. The prioritize-assumptions skill, defined in pm-product-discovery/skills/prioritize-assumptions/SKILL.md, evaluates each assumption on an Impact × Risk matrix and proposes the smallest possible validation experiment.

Execute this analysis using:

/prioritize-assumptions <<EOF
Assumption 1: Users will adopt a self-service analytics dashboard.
Assumption 2: Pricing tier will be a major driver of churn.
Assumption 3: Mobile reporting is a low-priority request.
EOF

The skill returns a prioritized matrix with concise experiment plans for high-impact, high-risk assumptions.

Step 3: Convert Insights to Prioritized Features

Transform validated assumptions into concrete feature ideas using the prioritize-features skill. As implemented in pm-product-discovery/skills/prioritize-features/SKILL.md, this applies ICE (Impact, Confidence, Ease) or RICE (Reach, Impact, Confidence, Effort) frameworks to score ideas by Impact, Effort, Risk, and Strategic Alignment.

Run the prioritization with:

/prioritize-features <<EOF
Feature A: Self-service analytics dashboard (allows users to build their own reports)
Feature B: Tier-based pricing and usage alerts
Feature C: Mobile-optimized reporting screen
Feature D: Export to PDF/Excel
Feature E: Integration with third-party BI tools
EOF

The output provides a ranked table of the top-five features with detailed rationale for each scoring dimension.

Step 4: Map Features to the Opportunity Solution Tree

Visualize where new work belongs in your product strategy using the opportunity-solution-tree skill. Located at pm-product-discovery/skills/opportunity-solution-tree/SKILL.md, this creates hierarchical diagrams linking validated opportunities to concrete solution branches.

Build your tree using:

/opportunity-solution-tree <<EOF
Opportunity 1: Lack of self-service analytics
  → Solution: Build a drag-and-drop dashboard builder
Opportunity 2: Pricing confusion leading to churn
  → Solution: Introduce tiered pricing with usage alerts
EOF

Step 5: Iterate Continuously

After shipping features or releasing prototypes, conduct fresh interviews and feed the new transcripts back into Step 1. This closed loop automatically surfaces new assumptions, re-prioritizes the backlog based on latest evidence, and prevents strategy drift.

Practical Implementation: Running the Complete Feedback Loop

All commands accept both inline text and file uploads via [upload <path>], enabling direct processing of CSV survey exports without copy-pasting.

Here is the complete iteration workflow as implemented in the phuryn/pm-skills environment:

# Process initial feedback

/interview summarize [upload ./customer-interview-march-15.txt]

# Extract and prioritize assumptions from the generated summary

/prioritize-assumptions [upload ./extracted-assumptions.md]

# Score resulting feature ideas

/prioritize-features <<EOF
Feature: Real-time collaboration dashboard
Feature: Automated weekly reporting
Feature: Role-based access controls
EOF

# Visualize strategic placement

/opportunity-solution-tree [upload ./top-opportunities.md]

# Post-launch: Process new feedback to validate assumptions

/interview summarize [upload ./validation-interviews.txt]

The commands referenced above—including /interview summarize detailed in pm-product-discovery/commands/interview.md—handle file parsing and markdown generation automatically, requiring no direct editing of skill files.

Summary

• pm-product-discovery structures workflows as reusable skills and executable commands, separating logic from interface.
• The summarize-interview skill transforms raw transcripts into structured JTBD and pain point summaries.
• prioritize-assumptions evaluates hypotheses on an Impact × Risk matrix and recommends validation experiments.
• prioritize-features applies ICE/RICE scoring to rank ideas by strategic alignment and effort.
• opportunity-solution-tree maps validated opportunities to concrete solution branches for roadmap visualization.
• All commands support file uploads and heredoc input, enabling pipeline integration with existing data sources.

Frequently Asked Questions

What is the difference between skills and commands in pm-product-discovery?

Skills are markdown specification files stored in pm-product-discovery/skills/ that define the logic, templates, and evaluation criteria for specific workflows. Commands are user-facing interfaces located in pm-product-discovery/commands/ that handle input parsing, file uploads, and workspace markdown creation while invoking the underlying skills. You interact with commands daily; you reference skill files when customizing behavior.

How does the prioritize-assumptions skill evaluate risk?

According to pm-product-discovery/skills/prioritize-assumptions/SKILL.md, the skill plots each assumption on an Impact × Risk matrix. High-impact, high-risk assumptions receive priority for validation, and the skill automatically proposes the smallest experiment capable of testing the assumption—such as a concierge test for workflow assumptions or a landing page test for demand validation.

Can I process batch feedback from multiple sources simultaneously?

Yes. All commands accept the [upload <path>] syntax, allowing you to pipe CSV exports from survey tools, support ticket databases, or interview transcripts directly into /prioritize-assumptions or /prioritize-features. The commands parse the input and generate individual markdown outputs for each processed item in your workspace.

How do I integrate the opportunity-solution-tree into my existing roadmap?

The opportunity-solution-tree skill outputs hierarchical markdown diagrams that link prioritized opportunities to solution branches. Import these diagrams into your roadmap tool or documentation platform as structured text. When new feedback arrives, rerun the tree generation with updated opportunities to visualize how shifting user needs affect your strategic solution map.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}