Foundation
Core Principles
Kanban is a change management method — not a framework. Start with what you do now.
🚦
Start with what you do now
No prescribed starting setup. Overlay Kanban on your existing process.
🔄
Evolutionary change
Small, continuous improvements over big transformations. Reduce resistance.
🙋
Leadership at all levels
Improvement is everyone's responsibility — not just management's.
🎯
Focus on customer needs
Understand what customers value and optimise delivery toward that.
🌊
Manage the work
Let people self-organise around the work. Don't manage the people.
🔍
Review regularly
Regularly review the network of services and update policies as needed.
Foundation
The Kanban Board
Each column is a workflow state. Cards move left to right. WIP limits are displayed per column.
Backlog
no limit
User login flow
Payment API
Dark mode
Ready
WIP: 3
🔥 Auth bug fix
User login flow
In Progress
WIP: 4
Dashboard redesign
🚫 CSV export
Email templates
Review
WIP: 2
Search indexing
Role permissions
Done
no limit
Onboarding flow
Rate limiting
Walk the board right-to-left in your daily meeting — starting from Done back to Backlog. This surfaces blockers closest to delivery first.
Practices
Six Core Practices
1 · Visualise the Workflow
Make all work visible — including blocked, waiting, and in-flight items. Hidden work is unmanageable work.
A good board includes: → All workflow states as columns → Work item types (colour-coded cards) → WIP limits displayed per column → Explicit policies per stage → Visible blockers and dependencies
2 · Limit WIP
The single most important Kanban practice. WIP limits enforce pull, reduce context switching, and surface bottlenecks immediately.
Setting WIP limits: Starting point = team members in that stage × 1.5 Observe → where does work pile up? Lower limits → faster flow, more discipline needed Little's Law → Cycle Time = WIP / Throughput
WIP limits are rules, not suggestions. When a limit is hit, the team stops starting and starts finishing — swarm on the blocked work before pulling anything new.
3 · Manage Flow
The goal is smooth, predictable flow — not maximum utilisation of people.
Key flow concepts: Pull system → work pulled when downstream capacity exists Bottleneck → stage where work accumulates; fix this first Little's Law → WIP = Throughput × Cycle Time Flow Efficiency = Active Time / Lead Time × 100%
Most teams have a flow efficiency of 15–40%. If yours is below 15%, there's a significant queue or wait-time problem worth investigating.
4 · Make Policies Explicit
Every decision about how work is handled should be written down and visible on the board.
Examples of explicit policies: → Entry/exit criteria for each column → Definition of Done per workflow stage → How blockers are escalated and resolved → Classes of service and their handling rules → How WIP limit exceptions are approved
5 & 6 · Feedback Loops & Improvement
Regular cadences create structured opportunities to inspect flow data and improve the system collaboratively.
Changes should be agreed by the team, grounded in flow data, and evolutionary — not imposed top-down. One small improvement per cadence beats quarterly big-bang changes.
Prioritisation
Classes of Service
Define how different types of work are prioritised and scheduled. Apply them at replenishment time.
🔥 Expedite
Business-critical. Bypasses WIP limits. Only one at a time.
e.g. production outage, security breach
📅 Fixed Date
Must arrive by a specific date. Risk managed proactively.
e.g. regulatory deadline, conference demo
📋 Standard
Normal business value. First-in, first-out within this class.
e.g. feature development, improvements
🌱 Intangible
Long-term value. No immediate urgency. Fills spare capacity.
e.g. tech debt, refactoring, research
Only one Expedite item should be in flight at a time. If everything is Expedite, nothing is — priorities become meaningless.
Prioritisation
Cadences
| Cadence | Frequency | Purpose |
|---|---|---|
| Strategy Review | Quarterly | Is the service fit for purpose? Adjust policies and goals. |
| Operations Review | Monthly | Review flow metrics across services. Address systemic issues. |
| Risk Review | Monthly | Identify and manage delivery risks proactively. |
| Service Delivery Review | Bi-weekly | Review service performance with stakeholders. |
| Replenishment Meeting | Weekly | Pull items into Ready. Prioritise by Class of Service. |
| Kanban Meeting | Daily | Walk the board right-to-left. Identify blockers. 15 minutes. |
| Delivery Planning | On demand | Coordinate release or deployment decisions. |
Measurement
Flow Metrics
Kanban teams use flow metrics, not velocity. These measure the system, not the people.
⏱
Cycle Time
Done Date − Start Date. How long items actually take once started.
Flow
📬
Lead Time
Done Date − Request Date. The customer's experience of wait.
Flow
📦
Throughput
Items completed per week or month. The team's delivery rate.
Flow
🔲
WIP
Items in progress at any moment. High WIP = high cycle time.
Flow
💧
Flow Efficiency
Active Time / Lead Time × 100%. % of time item is being worked on vs waiting.
Flow
Key formulas: Cycle Time = Done Date − Start Date Lead Time = Done Date − Request Date Throughput = Items Done / Time Period Flow Efficiency = Active Time / Lead Time × 100% Little's Law = WIP = Throughput × Cycle Time Benchmark: Flow Efficiency 15–40% typical · 40%+ high-performing
Measurement
Kanban vs Scrum
| Dimension | Kanban | Scrum |
|---|---|---|
| Cadence | Continuous flow | Fixed-length Sprints |
| Roles | No prescribed roles | Product Owner, Scrum Master, Developers |
| Change | Anytime | Between Sprints |
| WIP limits | Explicit, per column | Implicit (Sprint capacity) |
| Estimation | Optional | Usually required |
| Board | Persistent, evolving | Reset each Sprint |
| Best for | Ops, support, continuous delivery | Product development with clear goals |
Scrumban combines Sprint cadences with Kanban WIP limits and flow metrics. Common for teams transitioning from Scrum toward flow-based delivery.
Reference
Anti-Patterns
| Anti-Pattern | Problem | Fix |
|---|---|---|
| WIP limits as decoration | Team ignores limits when convenient | Limits are rules; violations must be discussed |
| No explicit policies | Inconsistent decisions across the team | Write entry/exit criteria for every column |
| Backlog as dumping ground | Everything goes in; nothing gets removed | Prune regularly; apply Class of Service |
| Everything is Expedite | Priorities become meaningless | One Expedite item at a time, maximum |
| Optimising for utilisation | People busy ≠ value delivered | Optimise for cycle time and throughput |
| Skipping the Kanban meeting | Blockers go unnoticed for days | Daily walk-the-board is non-negotiable |
| Too many columns | Complex board no one understands | Start with 4–5 columns; add only when needed |
| No flow metrics | No visibility into systemic issues | Track cycle time, throughput, WIP weekly |
Reference
Kanban Cheat Sheet
Six core practices 1. Visualise the workflow 2. Limit WIP 3. Manage flow 4. Make policies explicit 5. Implement feedback loops 6. Improve collaboratively Key metrics & formulas Lead Time = Done Date − Request Date Cycle Time = Done Date − Start Date Throughput = Items Done / Time Period Flow Efficiency = Active Time / Lead Time × 100% Little's Law = WIP = Throughput × Cycle Time Classes of service (priority order) 🔥 Expedite → now, bypass WIP limits, one at a time 📅 Fixed Date → schedule by date, manage risk 📋 Standard → FIFO, normal business value 🌱 Intangible → fill spare capacity WIP limit starting point Per active stage = team members × 1.5 Lower WIP → Lower Cycle Time (Little's Law) Flow Efficiency benchmark: 15–40% typical · 40%+ high-performing