KMP flashcards
KMP · Build recall with a focused review session.
32 cards. Read the term and the definition here. The deck at the top of this page is optional.
- Definition of Kanban
- A method for managing and improving knowledge work that visualizes the workflow, limits work in progress, manages flow, and makes process policies explicit, using an evolutionary rather than revolutionary change approach.
- Core Kanban practices
- Visualize the workflow, limit WIP, manage flow, make policies explicit, implement feedback loops, and improve collaboratively and experimentally. The first three are the minimum viable system.
- Work in progress limit
- A cap on the number of items allowed in a workflow state at one time. It exposes bottlenecks, forces completion before new starts, and shortens lead time by reducing multitasking.
- Little's Law
- Average WIP equals average throughput multiplied by average lead time. Knowing any two quantities determines the third, so increasing WIP at constant throughput necessarily lengthens lead time.
- Lead time versus cycle time
- Lead time runs from customer request to delivery; cycle time runs from work start to finish. Lead time includes waiting in queues before work begins.
- Pull system
- A system where downstream capacity signals authorize new work upstream. Work is pulled when a slot frees, rather than pushed into a queue regardless of downstream availability.
- Push system failure mode
- Work is assigned regardless of downstream capacity, so queues grow, lead time rises, and priority becomes unclear. The symptom is a board with full in-progress columns and stalled downstream states.
- Value stream mapping
- Walking the workflow backward from delivery to request to record every state and handoff a work item passes through, plus the wait time at each. It reveals queues invisible on the board.
- Classes of service
- Policies that differentiate handling of work items, such as expedite, fixed date, standard, and intangible. Each class has explicit selection and prioritization rules and its own capacity allocation.
- Expedite class policy
- A class for urgent items that bypass normal queuing. Policy must state who authorizes it, how many may be active simultaneously, and how much capacity it may consume.
- Ready queue column
- A board column holding items that are selected and eligible but not yet started. It buffers input and makes demand visible; it is not counted against active-state WIP limits.
- Definition of done per state
- Explicit exit criteria each item must satisfy before moving to the next column. Without it, items move prematurely and rework appears as unexpected flow in later states.
- Commitment point
- The moment a work item is accepted into the system and the clock for lead time starts. Before it, items are options being shaped, not commitments being delivered.
- Kanban board design rule
- Columns represent actual workflow states, not roles or people. Each active state carries a WIP limit, and the board should reflect the real handoffs work items experience.
- Setting an initial WIP limit
- Start from observed concurrent work in a state, reduce it modestly, and observe the effect. Limits are experiments to be tuned, not fixed targets derived from theory alone.
- Blocked item policy
- Explicit rules for flagging blocked work, who resolves it, and how long it may stay blocked. Blocked items still consume WIP capacity unless policy says otherwise.
- Flow efficiency
- Value-added time divided by total lead time. Because queue and wait time usually dominate, improving flow efficiency mainly means removing waiting, not accelerating active work.
- Cumulative flow diagram
- A stacked area chart of item counts per state over time. Band width shows queue size; a widening band signals accumulating work, a narrowing band signals draining.
- CFD widening band
- An expanding band between two states means arrivals to that state exceed departures. The response is to constrain upstream intake or add capacity, not to raise the downstream WIP limit.
- Cycle time control chart
- A run chart of individual item cycle times with statistical control limits. Points inside the limits are common-cause variation; points outside are special causes warranting investigation.
- Common versus special cause
- Common-cause variation is inherent in a stable process and is changed only by redesigning the system. Special-cause variation comes from assignable events and justifies targeted investigation.
- Throughput metric
- The number of work items completed per unit time. It is measured at the delivery point and pairs with WIP and lead time in Little's Law.
- Lead time distribution
- Reporting lead time as a percentile distribution rather than an average, so commitments can be made at a stated service level such as 85 percent of items within a given duration.
- Kanban cadences
- A set of recurring meetings at different time horizons: strategy review, operations review, risk review, service delivery review, replenishment, and daily standup. Each has a distinct decision purpose.
- Replenishment cadence
- The recurring meeting where items are selected from the backlog into the ready queue, respecting available capacity and class-of-service policies. It keeps intake aligned with throughput.
- Service delivery review
- A recurring review of whether the service meets expectations, using flow metrics such as lead time distribution, throughput, and blockers, and selecting improvement experiments to run.
- Operations review
- A cadence examining flow across multiple services or teams, resolving cross-team dependencies, and balancing capacity against demand at a level above the single board.
- Risk review
- A cadence focused on dependencies, blockers, and uncertain items, deciding how to address them before they disrupt flow. It complements the service delivery review's focus on metrics.
- Bottleneck identification
- The state where queues grow steadily and utilization is highest. Adding WIP upstream of it increases lead time without raising throughput; capacity must be added at the constraint.
- Improvement experiment loop
- State a hypothesis, change one policy or limit, measure flow metrics over a defined period, and keep or revert. This mirrors the scientific model of improvement in Kanban.
- Explicit policy examples
- Documented rules covering WIP limits, definition of ready and done, classes of service, prioritization, blocked-item handling, and escalation. Undocumented policy cannot be inspected or improved.
- Feedback loop hierarchy
- Loops run from short to long horizon: daily standup for coordination, replenishment for selection, delivery review for service, operations review for flow, risk review for dependencies, strategy review for portfolio.