Files
gh-lyndonkl-claude/skills/one-pager-prd/resources/methodology.md
2025-11-30 08:38:26 +08:00

13 KiB
Raw Blame History

One-Pager PRD Methodology

Table of Contents

  1. Problem Framing Techniques
  2. Metric Definition & Trees
  3. Scope Prioritization Methods
  4. Writing for Clarity
  5. Stakeholder Management
  6. User Story Mapping
  7. PRD Review & Iteration

1. Problem Framing Techniques

Jobs-to-be-Done Framework

Template: "When I [situation], I want to [motivation], so I can [outcome]."

Example: "Users want better search" → ✓ "When looking for a document among thousands, I want to filter by type/date/author, so I can find it in <30 seconds"

Application: Interview users for triggers, understand workarounds, quantify pain.

Problem Statement Formula

Template: "[User segment] struggles to [task] because [root cause], resulting in [quantified impact]. Evidence: [data]."

Example: "Power users (15% of users, 60% usage) struggle to edit multiple rows because single-row editing only, resulting in 5h/week manual work and 12% higher churn. Evidence: churn interviews (8/10 cited), analytics (300 edits/user/week)."

5 Whys (Root Cause Analysis)

Ask "why" 5 times to get from symptom to root cause.

Example: Users complain search is slow → Why? Query takes 3s → Why? DB not indexed → Why? Schema not designed for search → Why? Original use case was CRUD → Why? Product pivoted.

Root Cause: Product pivot created tech debt. Real problem: Re-architect for search-first, not just optimize queries.

Validation Checklist

  • Talked to 5+ users
  • Quantified frequency and severity
  • Identified workarounds
  • Confirmed top 3 pain point
  • Checked if competitors solve this
  • Estimated TAM

2. Metric Definition & Trees

SMART Metrics

Specific, Measurable, Achievable, Relevant, Time-bound.

"Improve engagement" → ✓ "Increase WAU from 10K to 15K (+50%) in Q2" "Make search faster" → ✓ "Reduce p95 latency from 3s to <1s by Q1 end"

Leading vs Lagging Metrics

Lagging: Outcomes (revenue, retention). Slow to change, final success measure. Leading: Early signals (adoption, usage). Fast to change, enable corrections.

Example: Lagging = Revenue. Leading = Adoption rate, usage frequency.

Metric Tree Decomposition

Break down North Star into actionable sub-metrics.

Example: Revenue = Users × Conversion × Price

  • Users = Signups + Reactivated
  • Conversion = Activate × Value × Paywall × Pay Rate

Application: Pick 2-3 metrics from tree you can influence.

Metric Anti-Patterns

Vanity: Signups without retention, page views without engagement Lagging-Only: Only revenue (too slow). Add adoption/engagement. Too Many: 15 metrics. Pick 1 primary + 2-3 secondary. No Baselines: "Increase engagement" vs "From 2.5 to 4 sessions/week"


3. Scope Prioritization Methods

MoSCoW Method

Must-Have: Feature doesn't work without this Should-Have: Important but feature works without it (next iteration) Could-Have: Nice to have if time permits Won't-Have: Explicitly out of scope

Example: Bulk Edit

  • Must: Select multiple rows, edit common field, apply
  • Should: Undo/redo, validation error handling
  • Could: Bulk delete, custom formulas
  • Won't: Conditional formatting (separate feature)

Kano Model

Basic Needs: If missing, users dissatisfied. If present, users neutral.

  • Example: Search must return results

Performance Needs: More is better (linear satisfaction)

  • Example: Faster search is better

Delight Needs: Unexpected features that wow users

  • Example: Search suggests related items

Application: Must-haves = Basic needs. Delights can wait for v2.

RICE Scoring

Reach: How many users per quarter? Impact: How much does it help each user? (0.25 = minimal, 3 = massive) Confidence: How sure are you? (0% to 100%) Effort: Person-months

Score = (Reach × Impact × Confidence) / Effort

Example:

  • Feature A: (1000 users × 3 impact × 80% confidence) / 2 months = 1200
  • Feature B: (5000 users × 0.5 impact × 100%) / 1 month = 2500
  • Prioritize Feature B

Value vs Effort Matrix

2×2 Grid:

  • High Value, Low Effort: Quick wins (do first)
  • High Value, High Effort: Strategic projects (plan carefully)
  • Low Value, Low Effort: Fill-ins (do if spare capacity)
  • Low Value, High Effort: Avoid

Determine Scope:

  1. List all potential features/requirements
  2. Plot on matrix
  3. MVP = High Value (both high/low effort)
  4. v2 = Medium Value
  5. Backlog/Never = Low Value

4. Writing for Clarity

Pyramid Principle

Structure: Lead with conclusion, then support.

Template:

  1. Conclusion: Main point (problem statement)
  2. Key Arguments: 3-5 supporting points
  3. Evidence: Data, examples, details

Example:

  1. Conclusion: "We should build bulk edit because power users churn without it"
  2. Arguments:
    • 12% higher churn among power users
    • Competitors all have bulk edit
    • Users hack workarounds (exports to Excel)
  3. Evidence: Churn analysis, competitive audit, user interviews

Application: Start PRD with executive summary using pyramid structure.

Active Voice & Concrete Language

Passive → Active:

  • "The feature will be implemented by engineering"
  • ✓ "Engineering will implement the feature"

Vague → Concrete:

  • "Improve search performance"
  • ✓ "Reduce search latency from 3s to <1s"

Abstract → Specific:

  • "Enhance user experience"
  • ✓ "Reduce clicks from 10 to 3 for common task"

Avoid Jargon & Acronyms

Bad: "Implement CQRS with event sourcing for the BFF layer to optimize TTFB for SSR"

Good: "Separate read and write databases to make pages load faster (3s → 1s)"

Rule: Define acronyms on first use, or avoid entirely if stakeholders unfamiliar.

Use Examples Liberally

Before (Abstract): "Users need flexible filtering options"

After (Concrete Example): "Users need flexible filtering. Example: Data analyst wants to see 'all invoices from Q4 2023 where amount > $10K and status = unpaid' without opening each invoice."

Benefits: Examples prevent misinterpretation, make requirements testable.

Scannable Formatting

Use:

  • Bullets: For lists (easier than paragraphs)
  • Headers: Break into sections (5-7 sections max per page)
  • Bold: For key terms (not entire sentences)
  • Tables: For comparisons or structured data
  • Short paragraphs: 3-5 sentences max

Avoid:

  • Long paragraphs (>6 sentences)
  • Wall of text
  • Too many indentation levels (max 2)

5. Stakeholder Management

Identifying Stakeholders

Categories:

  • Accountable: PM (you) - owns outcome
  • Approvers: Who must say yes (eng lead, design lead, exec sponsor)
  • Contributors: Who provide input (engineers, designers, sales, support)
  • Informed: Who need to know (marketing, legal, customer success)

Mapping:

  1. List everyone who touches this feature
  2. Categorize by role
  3. Identify decision-makers vs advisors
  4. Note what each stakeholder cares about most

Tailoring PRD for Stakeholders

For Engineering:

  • Emphasize: Technical constraints, dependencies, edge cases
  • Include: Non-functional requirements (performance, scalability, security)
  • Avoid: Over-specifying implementation

For Design:

  • Emphasize: User flows, personas, success criteria
  • Include: Empty states, error states, edge cases
  • Avoid: Specifying UI components

For Business (Sales/Marketing/Exec):

  • Emphasize: Business impact, competitive positioning, revenue potential
  • Include: Go-to-market plan, pricing implications
  • Avoid: Technical details

For Legal/Compliance:

  • Emphasize: Privacy, security, regulatory requirements
  • Include: Data handling, user consent, audit trails
  • Avoid: Underestimating compliance effort

Getting Buy-In

Technique 1: Pre-socialize

  • Share draft with key stakeholders 1:1 before group review
  • Gather feedback, address concerns early
  • Avoid surprises in group meeting

Technique 2: Address Objections Preemptively

  • Anticipate concerns (cost, timeline, technical risk)
  • Include "Risks & Mitigation" section
  • Show you've thought through trade-offs

Technique 3: Present Options

  • Don't present single solution as fait accompli
  • Offer 2-3 options with pros/cons
  • Recommend one, but let stakeholders weigh in

Technique 4: Quantify

  • Back up claims with data
  • "This will save users 5 hours/week" (not "improve efficiency")
  • Estimate revenue impact, churn reduction, support ticket decrease

6. User Story Mapping

Concept

Map user journey horizontally (steps), features vertically (priority).

Structure

Backbone (Top Row): High-level activities Walking Skeleton (Row 2): MVP user flow Additional Features (Rows 3+): Nice-to-haves

Example: E-commerce Checkout

Backbone:      Browse Products → Add to Cart → Checkout → Pay → Confirm
Walking Skeleton: View list    → Click "Add"  → Enter info → Card → Email receipt
v2:           Filters, search  → Save for later → Promo codes → Wallet → Order tracking
v3:           Recommendations  → Wish list    → Gift wrap  → BNPL → Returns

Application to PRD

  1. Define backbone: Key user activities (happy path)
  2. Identify MVP: Minimum to complete journey
  3. Slice vertically: Each slice is shippable increment
  4. Prioritize: Top rows are must-haves, bottom rows are v2/v3

Benefits

  • Visual representation of scope
  • Easy to see MVP vs future
  • Stakeholder alignment on priorities
  • Clear release plan

7. PRD Review & Iteration

Review Process

Phase 1: Self-Review

  • Let draft sit 24 hours
  • Re-read with fresh eyes
  • Check against rubric
  • Run spell check

Phase 2: Peer Review

  • Share with fellow PM or trusted colleague
  • Ask: "Is problem clear? Is solution sensible? Any gaps?"
  • Iterate based on feedback

Phase 3: Stakeholder Review

  • Share with approvers (eng, design, business)
  • Schedule review meeting or async comments
  • Focus on: Do we agree on problem? Do we agree on approach? Are we aligned on scope/metrics?

Phase 4: Sign-Off

  • Get explicit approval from each approver
  • Document who approved and when
  • Proceed to detailed design/development

Feedback Types

Clarifying Questions:

  • "What do you mean by X?"
  • Response: Clarify in PRD (you were unclear)

Concerns/Objections:

  • "This will take too long / cost too much / won't work because..."
  • Response: Address in Risks section or adjust approach

Alternative Proposals:

  • "What if we did Y instead?"
  • Response: Evaluate alternatives, update PRD if better option

Out of Scope:

  • "Can we also add feature Z?"
  • Response: Acknowledge, add to "Out of Scope" or "Future Versions"

Iterating Based on Feedback

Don't:

  • Defend original idea blindly
  • Take feedback personally
  • Ignore concerns

Do:

  • Thank reviewers for input
  • Evaluate objectively (is their concern valid?)
  • Update PRD with new information
  • Re-share revised version with change log

Version Control

Track Changes:

  • V1.0: Initial draft
  • V1.1: Updated based on eng feedback (added technical constraints)
  • V1.2: Updated based on design feedback (clarified user flows)
  • V2.0: Approved version

Change Log:

  • Date, version, what changed, why
  • Helps stakeholders see evolution
  • Prevents confusion ("I thought we agreed on X?")

Example:

## Change Log

**V1.2 (2024-01-15):**
- Updated scope: Removed bulk delete (security concern from eng)
- Added section: Performance requirements (feedback from eng)
- Clarified metric: Changed "engagement" to "% users who use bulk edit >3 times/week"

**V1.1 (2024-01-10):**
- Added personas: Broke out "power users" into data analysts vs operations teams
- Updated flows: Added error handling for validation failures

**V1.0 (2024-01-05):**
- Initial draft

When to Stop Iterating

Signs PRD is Ready:

  • All approvers have signed off
  • Open questions resolved or have clear decision timeline
  • Scope boundaries clear (everyone agrees on in/out)
  • Metrics defined and measurable
  • No blocking concerns remain

Don't:

  • Iterate forever seeking perfection
  • Delay unnecessarily
  • Overspecify (leave room for design/eng creativity)

Rule: If 80% aligned and no blocking concerns, ship it. Can iterate during development if needed.


Quick Reference: Methodology Selection

Use Jobs-to-be-Done when:

  • Framing new problem
  • Understanding user motivation
  • Validating problem-solution fit

Use Metric Trees when:

  • Defining success criteria
  • Connecting feature to business outcomes
  • Identifying leading indicators

Use MoSCoW/RICE when:

  • Prioritizing features for MVP
  • Managing scope creep
  • Communicating trade-offs

Use Pyramid Principle when:

  • Writing executive summary
  • Structuring arguments
  • Making PRD scannable

Use Stakeholder Mapping when:

  • Complex cross-functional initiative
  • Need to manage many approvers
  • Tailoring PRD for different audiences

Use Story Mapping when:

  • Planning multi-phase rollout
  • Aligning team on scope
  • Visualizing user journey

Use Review Process when:

  • Every PRD (always peer review)
  • High-stakes features (multiple review rounds)
  • Contentious decisions (pre-socialize)