Agile Project Management With Kanban

Introduction

• Kanban definition: Use a visual board with columns representing workflow stages
• Core benefit: Visualize work and bottlenecks to improve flow efficiency
• Agile enhancement: Apply Kanban to make any agile process more effective
• Simplicity focus: Start simple and evolve your process based on team needs
• Universal application: Implement Kanban in any development environment
• History note: Understand Kanban originated in Toyota manufacturing and was adapted for software
• Results orientation: Focus on delivering value, not following process dogma
• Expectations setting: Recognize that Kanban implementation takes practice and iteration

Getting Started with Kanban

• Start simple: Begin with a basic three-column board: To Do, Doing, Done
• Visualize all work: Put every work item on the board, including non-feature work
• Card essentials: Include title, owner, and estimated effort on each card
• Physical vs. digital: Choose board type based on team co-location and needs
• Team agreement: Get team buy-in on initial board design and rules
• Board placement: Make board highly visible to entire team
• Initial WIP limits: Start with simple limits like “number of team members minus one”
• Tracking creation: Track when items are created to measure lead time
• Quick wins: Look for immediate process problems revealed by visualization

Work Items

• Define work items: Create clear, manageable chunks of work
• Right-size items: Make items completable within a few days
• Item attributes: Include title, description, acceptance criteria, and estimate
• Item types: Differentiate between features, bugs, technical debt, and other work
• Estimation approach: Use simple T-shirt sizing or story points for estimates
• Splitting technique: Split large items vertically (thin slices of functionality)
• Dependencies: Clearly mark dependencies between items
• Definition of Ready: Ensure items are fully prepared before entering workflow
• Definition of Done: Define clear completion criteria for items

Workflow

• Map your process: Create columns that reflect actual workflow stages
• Keep it visible: Ensure entire workflow fits on one visible board
• Column policies: Define clear entry and exit criteria for each column
• Handoff clarity: Make handoff points between team members explicit
• Buffer columns: Add buffer columns before bottlenecks if needed
• Testing integration: Incorporate testing throughout the workflow, not just at end
• Done means done: Once in Done, items should need no further work
• Expedite lane: Create a separate lane for emergency items if needed
• Continuous improvement: Refine workflow visualization as understanding improves

Work in Progress (WIP) Limits

• Set WIP limits: Apply limits to workflow columns based on team capacity
• Purpose understanding: Recognize WIP limits exist to improve flow, not restrict team
• Start conservative: Begin with tighter limits and adjust as needed
• Column-specific limits: Set different limits for different workflow stages
• Respect the limits: Treat WIP limits as hard constraints, not suggestions
• Bottleneck detection: Use breached WIP limits to identify bottlenecks
• Team swarming: When a column reaches its limit, swarm to help clear it
• Adjustment process: Regularly review and adjust limits based on observed flow
• Personal WIP: Encourage team members to limit their personal work in progress

Metrics and Reporting

• Lead time: Measure time from request creation to completion
• Cycle time: Track time from work start to completion
• Throughput: Count items completed per unit time (week/month)
• Cumulative flow: Use cumulative flow diagrams to visualize workflow health
• WIP correlation: Notice correlation between high WIP and slower delivery
• Bottleneck identification: Use metrics to identify and address bottlenecks
• Predictability: Use historical cycle time to predict future delivery
• Improvement tracking: Measure effectiveness of process changes
• Simple approach: Start with basic metrics before adding complexity

Setting Expectations and Forecasting

• Probabilistic forecasting: Use cycle time distributions for realistic forecasts
• Avoid date commitments: Provide probability-based forecasts instead of firm dates
• Percentile thinking: Express forecasts as “85% likely to complete by X date”
• Monte Carlo simulations: Use historical data to simulate possible futures
• Scope management: Manage scope actively to achieve time goals
• Forecast updates: Update forecasts as work progresses and data improves
• Confidence explanation: Help stakeholders understand probabilistic forecasts
• Warning signs: Watch for early indicators that forecasts may be at risk
• Trust building: Use reliable forecasts to build stakeholder trust

Dealing with Meetings and Interruptions

• Meeting minimalism: Eliminate or streamline unnecessary meetings
• Stand-up format: Keep stand-ups focused on flow and blockers
• Interruption handling: Create explicit policies for handling interruptions
• Meeting preparation: Ensure meetings have clear agendas and goals
• Outside requests: Establish process for intake of requests from outside team
• Expedite policies: Create clear criteria for what can bypass normal flow
• Context switching: Measure and minimize harmful context switching
• Meeting rhythm: Establish consistent cadence for necessary meetings
• Continuous improvement: Make process improvement discussions regular events

Scaling Kanban

• Team of teams: Connect multiple team boards to visualize larger workflow
• Hierarchical items: Decompose large items into team-level work items
• Cross-team dependencies: Make dependencies across teams visible
• Escalation paths: Create clear paths for resolving cross-team blockers
• Consistent metrics: Use consistent metrics across teams for comparability
• Roll-up reporting: Aggregate metrics for program-level reporting
• Synchronized cadence: Consider synchronizing key meetings across teams
• WIP limits at scale: Apply WIP limits at program level as well as team level
• Improvement communities: Share Kanban improvements across teams

Advanced Topics

• Classes of service: Create different service classes for different work types
• Explicit policies: Document all process policies on or near the board
• Flow efficiency: Measure active time vs. waiting time for items
• Cost of delay: Consider prioritizing based on cost of delay
• Little’s Law: Understand the relationship between WIP, cycle time, and throughput
• Queue management: Actively manage queues to improve flow
• Slack allocation: Build in capacity for improvement and unexpected work
• System thinking: Optimize the entire system, not individual parts
• Upstream Kanban: Extend Kanban to idea generation and prioritization

Mixing Kanban with Other Methods

• Scrum+Kanban (Scrumban): Add visualization and WIP limits to Scrum
• Kanban for XP: Visualize flow while maintaining XP engineering practices
• Waterfall+Kanban: Improve traditional processes without full methodology switch
• Method pragmatism: Focus on principles and outcomes, not method purity
• Selective adoption: Take the best elements from different approaches
• Continuous delivery: Align Kanban with continuous delivery practices
• Process evolution: Start where you are and evolve toward better flow
• Cultural considerations: Adapt Kanban to work within existing culture
• Team empowerment: Let teams evolve their process based on experience

Sustaining and Improving Kanban

• Continuous improvement: Make regular process improvement habitual
• Metrics review: Regularly review metrics to identify improvement opportunities
• Retrospectives: Hold focused retrospectives on flow issues
• Flow experiments: Run controlled experiments to improve process
• Team ownership: Encourage team to own their process improvements
• Celebration: Recognize and celebrate process improvements
• Knowledge sharing: Create channels for sharing Kanban learnings
• Leadership support: Ensure continued leadership support for Kanban
• Value focus: Keep focus on delivery of value, not process compliance

Key Takeaways

1. Visualize work: Make all work visible on a board to reveal process issues
2. Limit WIP: Reduce work in progress to improve flow and quality
3. Manage flow: Optimize for smooth, predictable delivery of value
4. Make policies explicit: Document and display process rules
5. Implement feedback loops: Create mechanisms for regular improvement
6. Start where you are: Begin with current process and improve incrementally
7. Right-size work: Break work into small, manageable pieces
8. Data-driven decisions: Use metrics to guide process improvements
9. Probabilistic forecasting: Provide realistic date ranges based on historical data
10. Team empowerment: Let teams evolve their process based on real experience