Kanban | Agile Scrum Master
Kanban is a method for managing and improving flow by visualizing work, limiting work in progress, and evolving explicit policies through feedback. It helps teams deliver more predictably by reducing queues, exposing bottlenecks, and enabling continuous improvement of the end-to-end system. Key elements: workflow visualization, WIP limits, pull policies, classes of service, flow metrics such as lead time and throughput, and cadences for replenishment, delivery review, and retrospectives. Kanban can be applied to software, services, and other knowledge work.
How Kanban manages flow
Kanban is a method for managing and improving flow in knowledge work. It makes work visible, limits work in progress, and uses feedback loops to improve delivery capability. The objective is not to maximize utilization, but to improve end-to-end value delivery by reducing queues, delays, handoffs, and rework.
Kanban can start from the current way of working. Instead of a big reorganization, it evolves the system by making policies explicit and improving them based on observed performance. Kanban becomes effective when teams treat policies as hypotheses, measure flow, and change the system in small steps that can be inspected and adapted.
Core Principles of Kanban
- Start with what you do now - begin with the real workflow and improve what actually exists.
- Agree to pursue incremental, evolutionary change - improve in small steps so you can learn what works in your context.
- Respect current roles and responsibilities - evolve policies and collaboration without forcing a role redesign first.
- Encourage acts of leadership at all levels - make improvement everyone’s job, close to where the work happens.
Key Practices of Kanban
- Visualize the workflow - make work, queues, and waiting states visible, often on a Kanban Board, so you can inspect the system.
- Limit work in progress (WIP) - stop starting and start finishing to shorten feedback loops and reduce context switching.
- Manage flow - optimize for smooth movement of work and reliable delivery, not local busyness.
- Make policies explicit - define pull rules, priority rules, and quality expectations so decisions are clear and repeatable.
- Implement feedback loops - use cadences to inspect performance and adapt policies deliberately.
- Improve collaboratively, evolve experimentally - run small experiments, keep what improves outcomes, and remove what does not.
Kanban workflow design and WIP limits
Kanban design starts by mapping the real workflow from request to delivery, including waiting states and external dependencies. WIP limits expose constraints and force prioritization decisions. WIP limits are not only a board setting; they are a system-level agreement to reduce queues and protect finishing.
- Workflow states - define states that reflect real work and real waiting, including review, testing, and dependency steps.
- Pull criteria - specify when work may enter a state and what “ready to move” means to prevent hidden rework; these criteria should support flow, not become a heavy gate.
- WIP limits per state - cap active work where bottlenecks form to reduce delay and improve focus.
- Blocked work policies - make blocked items explicit and prioritize unblocking over starting new work.
- Classes of service - handle urgency explicitly so it does not silently override flow discipline.
As the system improves, teams adjust WIP limits and policies based on measured lead time, variability, and customer impact, not on preferences.
Kanban cadences and feedback loops
Kanban uses cadences to coordinate, inspect performance, and adapt policies without requiring timeboxed iterations. A cadence is useful when it produces decisions that change how work flows.
- Replenishment - decide what to pull next based on capacity, priority rules, and readiness criteria.
- Daily coordination - focus on flow: finishing, unblocking, and re-sequencing to protect delivery.
- Delivery planning - set delivery expectations using evidence from lead time and throughput variability.
- Service delivery review - inspect end-to-end performance and customer satisfaction signals.
- Improvement review - inspect policies, risks, and working agreements and change them deliberately; some teams use a retrospective format, but Kanban does not require a specific retrospective event.
When cadences become status reporting, learning stops. Keep cadences anchored in “what did we learn” and “what will we change.”
Kanban flow metrics and explicit policies
Kanban uses flow metrics to understand system capability and to forecast delivery. Metrics are for inspection and adaptation, not for ranking teams or pressuring output. The most useful metrics are those that lead to decisions about WIP, policies, and bottlenecks.
- Lead time - time from request to delivery, revealing customer responsiveness.
- Cycle time - time from start to finish, showing execution efficiency within the workflow.
- Throughput - completed items per period, useful for forecasting and capacity discussions.
- WIP and aging work in progress - leading indicators of delay, blocked work, and queue growth.
- Flow efficiency - ratio of active time to total time, highlighting waiting waste.
- Little's Law - helps reason about the relationship between WIP, throughput, and lead time when the system is sufficiently stable, supporting forecasting and policy decisions.
- Cumulative Flow Diagram (CFD) - visualizes WIP stability and bottlenecks over time.
Forecasting improves when work items are comparable in size, policies are stable enough to reduce variability, and forecasts are expressed as ranges rather than single-number promises.
Kanban adoption steps
Kanban can be adopted incrementally by starting with transparency and then strengthening flow discipline. The aim is to improve delivery capability and learning, not to perfect the board configuration.
- Map the current workflow - visualize real states, queues, and dependencies on a Kanban Board to establish shared reality.
- Define explicit policies - agree pull rules, priority rules, and quality criteria so decisions are consistent.
- Introduce WIP limits - start where bottlenecks exist and adjust based on measured effects.
- Measure flow - establish lead time, cycle time, throughput, and aging baselines.
- Improve iteratively - run small experiments, review results, and evolve policies collaboratively.
Common misuse and practical guardrails
Kanban is often reduced to a visualization tool. When WIP discipline and explicit policies are missing, the board becomes a map of overload rather than a method for improving flow.
- Board without WIP limits - work keeps starting, queues grow, and delivery becomes unpredictable.
- Hidden blocked work - blockers stay invisible, aging increases, and finishing is delayed.
- Priority thrashing - constant reordering increases context switching and prevents learning from stable policies.
- Metrics as pressure - teams game numbers, hide problems, and reduce transparency.
- Local optimization - parts of the system improve while end-to-end lead time stays high due to handoffs.
- Definition of Ready as a gate - using a heavy readiness gate to delay entry into the workflow hurts by increasing queues and slowing learning, do instead by using lightweight pull criteria and explicit policies that support flow.
When these problems appear, return to fundamentals: make policies explicit, limit WIP, prioritize unblocking and finishing, and use flow metrics to guide improvements across the full value stream.
Kanban is a method for improving flow by visualizing work, limiting WIP, managing policies, and continuously optimizing the end-to-end delivery system