Implementation Velocity Tracking

Implementation Velocity Tracking

Tracks actual implementation velocity and provides predictive estimates for remaining work based on historical data.

Quick Start

Record wave completion metrics:

"Record metrics for wave-5.1.1: 28 actual hours, 24 estimated, 3 defects, 92% coverage"

Estimate remaining work:

"Estimate completion time for remaining waves in Feature 5 using historical velocity"

Analyze quality trends:

"Analyze defect density and test coverage trends for last 30 days"

Core Workflow

Step 1: Record Wave Completion Metrics

After each wave completes, record actual metrics:

Timing Metrics:

• Estimated hours (from wave plan)
• Actual hours worked
• Start date
• Completion date
• Duration (calendar days)

Quality Metrics:

• Defects found (during development)
• Defects found (post-release)
• Test coverage percentage
• Code review cycle time
• Rework percentage

Deliverables:

• User stories completed
• User stories deferred
• Acceptance criteria met
• Technical debt created

Step 2: Calculate Velocity Metrics

Wave Velocity:

Wave Velocity = Actual Hours / Estimated Hours

Examples:

• 1.0 = Perfect estimate
• 0.8 = Finished early (20% faster than estimated)
• 1.25 = Took longer (25% over estimate)

Rolling Average Velocity (last N waves):

Team Velocity = Average(Wave Velocities)

Velocity Trend:

• Improving: Velocity decreasing toward 1.0
• Stable: Velocity consistent
• Worsening: Velocity increasing (taking longer than estimated)

Step 3: Track Quality Metrics

Defect Density:

Defect Density = Defects Found / KLOC (thousands of lines of code)

Or simpler:

Defects per Wave = Average defects across waves

Test Coverage Trend:

• Track coverage percentage over time
• Identify waves with low coverage
• Set coverage targets (e.g., >90%)

Rework Percentage:

Rework % = (Rework Hours / Total Hours) × 100

Step 4: Generate Predictive Estimates

Estimate Remaining Work:

Adjusted Estimate = Base Estimate × Team Velocity

Example:

• Remaining waves: 5 waves
• Estimated hours: 120 hours total
• Team velocity: 1.12
• Adjusted estimate: 120 × 1.12 = 134 hours

Calendar Estimate (with team size):

Calendar Days = Adjusted Hours / (Team Size × Hours per Day)

Example:

• Adjusted hours: 134
• Team size: 2 developers
• Hours per day: 8
• Calendar days: 134 / (2 × 8) = 8.4 days

Confidence Level:

• High: Based on 15+ completed waves
• Medium: Based on 5-14 completed waves
• Low: Based on <5 completed waves

Step 5: Analyze Trends and Generate Reports

Create report in Docs/reports/velocity/metrics-{date}.md:

# Velocity and Quality Metrics Report

**Date**: YYYY-MM-DD
**Scope**: [Epic/Feature/All]
**Period**: Last 30 days

## Executive Summary
[1-2 sentences on team velocity and quality trends]

## Velocity Metrics
- Current Team Velocity: X.XX
- Trend: [Improving/Stable/Worsening]
- Completed Waves: X
- Average Hours per Wave: XX

## Quality Metrics
- Defect Density: X.X per wave
- Test Coverage: XX%
- Rework Percentage: XX%

## Predictive Estimates
- Remaining Waves: X
- Base Estimate: XX hours
- Adjusted Estimate: XX hours (with velocity)
- Expected Completion: YYYY-MM-DD

## Trend Analysis
[Charts and analysis of trends over time]

## Recommendations
[Actions to improve velocity or quality]

Use scripts/calculate_velocity.py and scripts/predict_completion.py.

Key Concepts

Velocity: Ratio of actual hours to estimated hours (1.0 = perfect estimate)

Team Velocity: Rolling average velocity across recent waves

Defect Density: Number of defects per wave or per KLOC

Test Coverage: Percentage of code covered by tests

Rework: Time spent fixing issues in already-completed code

Adjusted Estimate: Base estimate multiplied by team velocity

Confidence Level: Reliability of prediction based on historical data quantity

Velocity Trend: Direction velocity is moving (improving/stable/worsening)

Available Resources

Scripts

• scripts/calculate_velocity.py — Calculates velocity metrics from wave data

  python scripts/calculate_velocity.py \
    --wave wave-5.1.1 \
    --estimated 24 \
    --actual 28 \
    --output velocity-db.json
  # Output: Wave Velocity: 1.17, Team Velocity: 1.12
  

• scripts/predict_completion.py — Predicts completion date using historical velocity

  python scripts/predict_completion.py \
    --remaining-waves 5 \
    --estimated-hours 120 \
    --velocity-db velocity-db.json \
    --team-size 2
  # Output: Expected Completion: 2025-02-05 (±2 days)
  

• scripts/analyze_trends.py — Analyzes quality and velocity trends

  python scripts/analyze_trends.py \
    --velocity-db velocity-db.json \
    --metrics defect-density,coverage,velocity \
    --period 30
  # Output: Trend charts and analysis
  

References

• references/velocity-tracking-guide.md — Comprehensive guide to velocity tracking
• references/metrics-definitions.md — Definitions of all tracked metrics
• references/estimation-best-practices.md — How to improve estimation accuracy

Tracking Phases

Phase 1: Initial Calibration (First 3-5 Waves)

When: Starting new project or team Focus: Establish baseline velocity Output: Initial velocity estimate Note: Low confidence, wide variance expected

Phase 2: Velocity Stabilization (Waves 6-15)

When: Team finding rhythm Focus: Track velocity convergence Output: Medium confidence estimates Note: Velocity should stabilize around consistent value

Phase 3: Mature Velocity Tracking (Wave 15+)

When: Established team and process Focus: Fine-tuning and quality optimization Output: High confidence estimates Note: Focus shifts from velocity to quality metrics

Phase 4: Continuous Improvement

When: Ongoing Focus: Process optimization Output: Trend analysis and recommendations Note: Use data to drive process improvements

Common Velocity Patterns

Pattern 1: Early Optimism (First 3 waves)

Velocity starts low (0.6-0.8), then increases to 1.2-1.5
Cause: Initial estimates too optimistic
Solution: Calibrate estimates upward

Pattern 2: Learning Curve (Waves 1-10)

Velocity high initially (1.5+), gradually decreases to 1.0-1.1
Cause: Team learning codebase and process
Solution: Normal, velocity will stabilize

Pattern 3: Technical Debt Impact

Velocity gradually increasing (1.0 → 1.3 → 1.5)
Cause: Accumulated technical debt slowing development
Solution: Dedicate waves to refactoring

Pattern 4: Scope Creep

Velocity consistently high (1.3-1.5)
Cause: Scope creeping beyond estimates
Solution: Better scope control in wave planning

Pattern 5: Quality Issues

Velocity low (0.8-0.9) but defect density high
Cause: Rushing to meet deadlines, creating defects
Solution: Allow realistic timelines, improve quality

Output Format

Console Output:

Implementation Velocity Tracking
================================

Wave: 5.1.1 - Singleton Patterns
Completed: 2025-01-21

Velocity Metrics:
  Estimated Hours: 24
  Actual Hours: 28
  Wave Velocity: 1.17 (17% over estimate)

Team Velocity (last 8 waves):
  Average: 1.12
  Trend: ↓ Improving (was 1.25)
  Standard Deviation: 0.08

Quality Metrics:
  Defects Found: 3
  Defect Density: 2.8 per wave (↓ improving)
  Test Coverage: 92% (↑ improving)
  Rework: 2 hours (7%)

Updated Velocity Database: velocity-db.json

Predictive Estimate (remaining 5 waves):
  Base Estimate: 120 hours
  Adjusted (velocity 1.12): 134 hours
  With 2-person team: 67 hours/person
  Expected Completion: 2025-02-05 (±2 days)
  Confidence: MEDIUM (8 completed waves)

Report: Docs/reports/velocity/wave-5.1.1-metrics-2025-01-21.md

Integration with Workflow

/implement-waves integration:

## Step 7: Record Velocity Metrics

After wave completion:
- Record actual hours worked
- Record defects found
- Record test coverage achieved
- Invoke `implementation-velocity-tracking` skill
- Update velocity database
- Review velocity trends

/design-waves integration:

## Step 0.5: Adjust Estimates Using Velocity

Before estimating new waves:
- Check current team velocity
- Adjust estimates by velocity factor
- Use historical data for similar wave types
- Account for team size changes

/review-waves integration:

## Wave Retrospective: Velocity Review

During wave retrospective:
- Review velocity for completed wave
- Compare to team average
- Identify causes of variance
- Document lessons learned
- Adjust future estimates

Epic Planning:

## Epic Estimation

Before committing to epic:
- Estimate total hours for all features
- Multiply by team velocity
- Add buffer for uncertainty
- Calculate expected duration
- Get stakeholder buy-in on timeline

Success Criteria

• ✅ All completed waves have recorded metrics
• ✅ Velocity calculated accurately
• ✅ Quality metrics tracked consistently
• ✅ Trends analyzed and visualized
• ✅ Predictive estimates generated
• ✅ Confidence level appropriate
• ✅ Report generated in standard format
• ✅ Recommendations provided for improvement

Tips for Accurate Velocity Tracking

1. Record immediately - Record metrics right after wave completion
2. Be honest - Accurate data is more valuable than optimistic data
3. Track consistently - Use same metrics for all waves
4. Sufficient history - Need 5+ waves for reliable predictions
5. Adjust for context - Account for team changes, complexity variations
6. Review trends - Look for patterns, not individual waves
7. Use for learning - Velocity is a tool for improvement, not judgment

Examples

Example 1: Record Wave Completion

Command:

"Record velocity metrics for wave-5.1.1: 28 actual hours, 24 estimated, 3 defects, 92% coverage"

Process:

1. Read velocity database (velocity-db.json)
2. Calculate wave velocity: 28 / 24 = 1.17
3. Update rolling average: (previous avg + 1.17) / n
4. Record quality metrics
5. Save to database

Output:

• Wave Velocity: 1.17 (17% over)
• Team Velocity: 1.12 (updated average)
• Defect Density: 2.8 per wave (improving)
• Test Coverage: 92% (on target)

Example 2: Predict Remaining Work

Command:

"Estimate completion for remaining Feature 5 waves using historical velocity"

Process:

1. Read velocity database
2. Get team velocity: 1.12
3. Find remaining waves: 5 waves
4. Sum estimated hours: 120 hours
5. Apply velocity: 120 × 1.12 = 134 hours
6. Calculate calendar: 134 / (2 × 8) = 8.4 days
7. Add confidence interval: ±2 days

Output:

• Adjusted Estimate: 134 hours
• Expected Completion: 2025-02-05
• Confidence: MEDIUM (8 waves of history)

Example 3: Analyze Quality Trends

Command:

"Analyze defect density and test coverage trends for last 30 days"

Process:

1. Read velocity database
2. Filter waves from last 30 days
3. Calculate defect density trend
4. Calculate test coverage trend
5. Identify anomalies
6. Generate recommendations

Output:

Quality Trends (Last 30 Days):

Defect Density:
  Trend: 📈 IMPROVING (4.2 → 2.8 per wave)
  Target: <3.0 ✅ ACHIEVED

Test Coverage:
  Trend: 📈 IMPROVING (85% → 92%)
  Target: >90% ✅ ACHIEVED

Alert: Defect spike in wave-5.3.1 (7 defects)
  Recommendation: Investigate root cause

Last Updated: 2025-01-21