Agent skill
tradeoff-decision
Use when A/B tests show mixed results or features have conflicting metrics - evaluates net positive/negative through strategic alignment, explores mitigation strategies before binary rollback decisions
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/data/tradeoff-decision
SKILL.md
Tradeoff Decision Workflow
Purpose
Make informed ship/no-ship/iterate decisions when A/B test results are mixed (some metrics up, others down) or when a launched feature shows conflicting performance. Avoids binary thinking by exploring nuanced approaches that preserve gains while addressing losses.
When to Use This Workflow
Use this workflow when:
- A/B test shows some metrics increasing, others decreasing
- Feature launched with mixed performance across different metrics
- Stakeholders disagree on whether to ship or rollback
- Redesign shifted user behavior in unexpected ways
- Need to choose between two good but mutually exclusive options
- Optimizing for one metric clearly harms another
Skills Sequence
This workflow applies four skills systematically:
1. North Star Alignment
↓ (Determine which metric matters more strategically)
2. Trade-off Evaluation (Short vs. Long-term)
↓ (Assess temporal trade-offs)
3. Trade-off Evaluation (Mitigation)
↓ (Explore "best of both worlds" approaches)
4. Funnel-Based Metric Mapping
↓ (Identify where in user journey problem occurs)
OUTPUT: Assessment (net positive/negative/depends),
Recommended action (ship/rollback/iterate),
Mitigation strategies, Risks
Required Inputs
Gather this information before starting:
Metric Changes
-
Metrics that increased
- Which metrics went up?
- By how much (absolute and %)?
- Statistical significance?
-
Metrics that decreased
- Which metrics went down?
- By how much (absolute and %)?
- Statistical significance?
Context
-
Change description
- What was tested or launched?
- What was the hypothesis?
- How long has it been running?
-
Company strategic priorities
- What's most important to company now?
- Growth, monetization, retention, competitive positioning?
-
User journey context
- Where in the funnel does change occur?
- Which user behaviors shifted?
Current State
-
Rollback feasibility
- Can we easily revert?
- What's the cost of reverting?
- How many users affected?
-
Iteration options
- Can we modify the change?
- What variations could we test?
- Timeline for iteration?
Workflow Steps
Step 1: Strategic Alignment Assessment (15 minutes)
Use the north-star-alignment skill
Determine which metric matters more to company strategy:
Questions to answer:
-
Which metrics align with North Star?
- Do positive metrics connect to company-level goals?
- Do negative metrics impact North Star metrics?
-
Which metrics align with mission?
- Does change serve company mission?
- Does it strengthen core value proposition?
-
Which metrics drive long-term competitive advantage?
- Does change improve differentiation?
- Does it strengthen market position?
-
Which metrics feed important flywheels?
- Do positive metrics create virtuous cycles?
- Do negative metrics break retention loops?
Output:
## Strategic Alignment Assessment
**Metrics Aligned with Strategy:**
- [Metric that went up/down]: [Why strategically important]
- [Metric that went up/down]: [Why strategically important]
**Strategic Priority Ranking:**
1. [Most important metric]: [Rationale]
2. [Second priority]: [Rationale]
3. [Lower priority]: [Rationale]
**Mission Alignment:**
- Change serves mission: [Yes/No/Partially]
- Rationale: [Explanation]
**Competitive Positioning:**
- Strengthens positioning: [Yes/No/Mixed]
- Rationale: [Explanation]
Step 2: Temporal Trade-off Analysis (15 minutes)
Use the tradeoff-evaluation skill - Short vs. Long-term Framework
Evaluate whether this is short-term gain/long-term loss or vice versa:
Framework questions:
-
Short-term Impact (this quarter)
- Immediate revenue effect?
- This quarter's growth?
- Current user satisfaction?
-
Long-term Impact (12+ months)
- User retention trajectory?
- Flywheel effects?
- Sustainable competitive advantage?
- Brand and market position?
-
Irreversibility Assessment
- Can we reverse course later if needed?
- What becomes locked in?
- What habits are we building?
Apply Snapchat test:
- Does this improve vanity metrics (time-on-site) while breaking engagement flywheel (messages/stories)?
- Are we optimizing for wrong thing?
Output:
## Temporal Analysis
**Short-term (0-3 months):**
- Revenue impact: [Positive/Negative/Neutral - magnitude]
- Growth impact: [Positive/Negative/Neutral - magnitude]
- User satisfaction: [Positive/Negative/Neutral]
**Long-term (12+ months):**
- Retention impact: [Projected effect]
- Flywheel effects: [Strengthened/Weakened/Neutral]
- Competitive position: [Stronger/Weaker/Same]
- Strategic alignment: [High/Medium/Low]
**Trade-off Classification:**
- [Short-term gain, long-term loss] → Risky
- [Short-term loss, long-term gain] → Consider if affordable
- [Short-term gain, long-term gain] → Ship
- [Short-term loss, long-term loss] → Don't ship
**Recommendation based on temporal analysis:**
[Ship/Don't ship/Depends on mitigation]
Step 3: Mitigation Strategy Exploration (20-30 minutes)
Use the tradeoff-evaluation skill - Mitigation Framework
Before deciding to rollback or ship as-is, explore "best of both worlds" options:
Mitigation strategy categories:
Strategy 1: Segmented Rollout
When to use: Impact varies by user segment
Analysis:
- Which segments show net positive?
- Which segments show net negative?
- Can we ship to positive segments only?
Implementation:
- Define segment criteria
- Rollout plan for positive segments
- Iteration plan for negative segments
Strategy 2: Feature Modification
When to use: Specific element causing harm
Analysis:
- Which specific change caused negative impact?
- Can we keep beneficial parts, modify harmful parts?
- What variations could preserve gains while fixing losses?
Implementation:
- Identify problematic component
- Design modified version
- A/B test: Original vs. Modified vs. Current
Strategy 3: Compensation Mechanisms
When to use: Trade-off is necessary but painful
Analysis:
- What complementary features could compensate?
- Can we create new paths to value?
- Can we boost other areas to offset loss?
Implementation:
- Identify compensation opportunities
- Parallel feature development
- Monitor net effect
Strategy 4: Gradual Rollout with Monitoring
When to use: Uncertain about long-term effects
Analysis:
- What could we learn from gradual rollout?
- What delayed effects might emerge?
- What monitoring would de-risk?
Implementation:
- 10% → 25% → 50% over weeks
- Maintain holdback group
- Set rollback trigger criteria
Strategy 5: Threshold-Based Approach
When to use: Acceptable within certain bounds
Analysis:
- What magnitude of negative impact is acceptable?
- What thresholds trigger concerns?
- Can we monitor and revert if crossed?
Implementation:
- Define acceptable thresholds
- Set up automated alerts
- Establish rollback criteria
Output:
## Mitigation Strategies
### Evaluated Strategies:
**Strategy 1: [Name]**
- Approach: [Description]
- Preserves: [Positive metrics]
- Addresses: [Negative metrics]
- Feasibility: [Easy/Medium/Hard]
- Timeline: [How long to implement]
- Risks: [What could go wrong]
**Strategy 2: [Name]**
[Same structure]
**Strategy 3: [Name]**
[Same structure]
### Recommended Mitigation:
**[Strategy name]**
- Why: [Rationale]
- Implementation: [Specific steps]
- Success criteria: [How to know it worked]
- Fallback plan: [If mitigation fails]
Step 4: Funnel Location Analysis (10 minutes)
Use the funnel-metric-mapping skill
Identify where in the user journey the trade-off occurs:
Questions:
-
Which funnel stage is affected?
- Reach, Activation, Engagement, Retention?
- Multiple stages?
-
Is impact concentrated or distributed?
- Single point of friction?
- Systemic shift across journey?
-
Do metrics at different stages conflict?
- Earlier stage improved, later stage harmed?
- Suggests optimization for wrong outcome
-
What transitions are affected?
- Conversion rates between stages changed?
- Drop-off points shifted?
Output:
## Funnel Impact Analysis
**Affected stages:**
- [Stage name]: [Metric change description]
- [Stage name]: [Metric change description]
**Transition impacts:**
- [Stage A → Stage B]: [Conversion rate change]
**Concentration:**
- [Single point / Distributed across journey]
**Insights:**
- [What funnel analysis reveals about trade-off]
Step 5: Make Recommendation (10 minutes)
Synthesize all analyses into clear recommendation:
Decision options:
-
Ship Fully
- When: Net positive across segments, strategic alignment high
- Action: Roll out to 100%
- Monitoring: Track counter-metrics
-
Ship to Segments
- When: Positive for some users, negative for others
- Action: Roll out to net-positive segments
- Plan: Iterate for problematic segments
-
Iterate First
- When: Promising but needs modification
- Action: Implement mitigation strategy
- Timeline: Test modified version before full ship
-
Gradual Rollout
- When: Uncertain about long-term effects
- Action: Staged rollout with monitoring
- Triggers: Rollback criteria defined
-
Rollback
- When: Net negative, no clear mitigation, high risk
- Action: Revert to previous state
- Learning: Document why for future
Recommendation document:
# Trade-off Decision: [Change Name]
## Executive Summary
[One paragraph: Metrics moved, strategic assessment, recommendation]
## Metric Changes
**Positive:**
- [Metric]: +X% ([why this matters])
- [Metric]: +X% ([why this matters])
**Negative:**
- [Metric]: -X% ([why this matters])
- [Metric]: -X% ([why this matters])
## Strategic Assessment (Step 1)
**Most important metric:** [Name]
**Rationale:** [Why strategically critical]
**Alignment:** [High/Medium/Low]
## Temporal Analysis (Step 2)
**Short-term:** [Net positive/negative]
**Long-term:** [Net positive/negative]
**Trade-off type:** [Classification]
## Mitigation Options (Step 3)
**Explored:**
1. [Strategy name]: [Feasibility]
2. [Strategy name]: [Feasibility]
**Recommended:** [Strategy name]
## Funnel Impact (Step 4)
**Affected stage:** [Stage name]
**Insight:** [What this reveals]
## Recommendation
**Decision:** [Ship Fully / Ship to Segments / Iterate First / Gradual / Rollback]
**Rationale:**
- [Point 1]
- [Point 2]
- [Point 3]
**Implementation:**
- Action: [Specific steps]
- Owner: [Name/team]
- Timeline: [When]
- Success criteria: [How to measure]
**Risks:**
- Risk 1: [Description + mitigation]
- Risk 2: [Description + mitigation]
**Monitoring Plan:**
- Metrics to track: [List]
- Review frequency: [Daily/Weekly]
- Rollback triggers: [Conditions that require revert]
- Responsible: [Name/team]
## Appendices
- A/B test results: [Detailed data]
- Segment analysis: [Breakdown]
- Stakeholder input: [Summary]
Decision Matrix
Use this matrix for quick assessment:
| Strategic Alignment | Short-term | Long-term | Mitigation Feasible | Recommendation |
|---|---|---|---|---|
| High | + | + | N/A | Ship Fully |
| High | + | - | Yes | Iterate First |
| High | + | - | No | Gradual with Monitoring |
| High | - | + | N/A | Ship if affordable |
| Low | + | - | Any | Rollback |
| Low | - | - | Any | Rollback |
| Mixed | Mixed | + | Yes | Iterate First |
| Mixed | Mixed | - | No | Rollback |
Common Mistakes
| Mistake | Fix |
|---|---|
| Binary thinking (ship or rollback only) | Explore mitigation strategies first |
| Not assessing strategic alignment | Start with North Star alignment |
| Ignoring long-term flywheel effects | Complete temporal analysis |
| Immediate rollback without diagnosis | Use metric-diagnosis first if root cause unclear |
| Treating all metrics as equal | Rank metrics by strategic importance |
| Not considering segmentation | Analyze by user segment |
Success Criteria
Tradeoff decision succeeds when:
- Strategic alignment assessed (which metric matters more)
- Temporal trade-offs evaluated (short vs. long-term)
- 3-5 mitigation strategies explored
- Funnel impact analyzed
- Clear recommendation made with rationale
- Implementation plan defined
- Risks identified with mitigation plans
- Monitoring plan established
- Stakeholders aligned on decision
Real-World Example: Snapchat Redesign
Context
- Redesign launched: Separate social from media
- Time on site: ↑ 15%
- Messages sent: ↓ 12%
- Stories shared: ↓ 8%
Step 1: Strategic Alignment (15 min)
Questions:
- Which metric aligns with Snapchat's strategic positioning?
- Camera/social app or content consumption app?
Analysis:
-
Content app positioning (time-on-site):
- Market saturated (TikTok, YouTube, Instagram)
- Not core competency
- Short-term thinking
-
Social/camera app positioning (messages/stories):
- Unique value proposition
- Core competency
- Sustainable competitive advantage
Conclusion: Messages and stories strategically more important than time-on-site
Step 2: Temporal Analysis (15 min)
Short-term (this quarter):
- Revenue: Positive (more ads shown with higher time-on-site)
- Growth: Neutral
- User satisfaction: Mixed (some like creators, some miss social)
Long-term (12 months):
- Retention: Negative risk (broken social flywheel)
- Fewer stories → Fewer notifications → Less app opening
- Fewer messages → Less reciprocal engagement
- Flywheel: Weakened critical loop
- Competitive position: Confused (competing with TikTok in saturated market)
Classification: Short-term gain (time/ads), long-term loss (retention/flywheel)
Assessment: Net negative
Step 3: Mitigation Strategies (25 min)
Strategy 1: Immediate Rollback
- Pros: Restores social flywheel quickly
- Cons: Loses time-on-site gains, admits failure
- Feasibility: Easy
Strategy 2: Targeted Modifications
- Approach: Keep creator content but re-integrate friend stories
- Implementation:
- Show friend stories in creator tab
- Add "message friend" prompts after content viewing
- Balance algorithm between creator and friend content
- Preserves: Time-on-site gains
- Addresses: Social engagement losses
- Feasibility: Medium complexity, 4-6 weeks
Strategy 3: Segmented Experience
- Heavy social users: Friend-focused interface
- Content consumers: Creator-focused interface
- Feasibility: Complex, 8-12 weeks
Recommended: Strategy 2 (Targeted Modifications)
- Best balance of preserving gains and fixing losses
- Reasonable timeline
- Testable before full rollout
Step 4: Funnel Analysis (10 min)
Affected stages:
- Engagement (Depth): Messages and stories down (core actions)
- Engagement (Breadth): Time on site up (passive consumption)
Insight:
- Optimizing breadth at expense of depth
- Depth drives retention flywheel, breadth doesn't
- Classic mistake: Vanity metric over strategic metric
Step 5: Recommendation (10 min)
Decision: Iterate First (Don't immediately rollback, don't keep as-is)
Rationale:
- Clear strategic importance of social metrics
- Long-term retention risk outweighs short-term revenue gain
- Feasible mitigation strategy exists
- Can preserve some gains while fixing critical issue
Implementation:
- Action: Implement targeted modifications (Strategy 2)
- Owner: Product team + Design
- Timeline: 4-6 weeks for modified version
- Success criteria:
- Messages sent: Recover to within 5% of baseline
- Stories shared: Recover to within 5% of baseline
- Time on site: Maintain at least 50% of gains
Risks:
- Risk: Modifications don't fix social engagement
- Mitigation: Full rollback if metrics don't improve in 30 days
- Risk: Lose all time-on-site gains
- Mitigation: Acceptable trade-off for strategic alignment
Monitoring:
- Daily: Messages, stories, time-on-site
- Weekly: Full funnel review
- Rollback trigger: If messages/stories drop further 5%
Timeline: 90 minutes for full analysis
Related Skills
This workflow orchestrates these skills:
- north-star-alignment (Step 1)
- tradeoff-evaluation (Steps 2-3)
- funnel-metric-mapping (Step 4)
Related Workflows
- metric-diagnosis: Use first if root cause of metric changes unclear
- metrics-definition: Defines counter-metrics to detect trade-offs early
- dashboard-design: Monitors trade-offs through balanced metric sets
Time Estimate
Total: 70-90 minutes
- Step 1 (Strategic alignment): 15 min
- Step 2 (Temporal analysis): 15 min
- Step 3 (Mitigation): 20-30 min
- Step 4 (Funnel analysis): 10 min
- Step 5 (Recommendation): 10 min
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?