Value Stream Mapping and Continuous Improvement
Value Stream Mapping and Continuous Improvement
This domain tests whether you can map how work actually flows to a customer, quantify the delay, and target improvement at the constraint.
Map mechanics
A value stream map shows process steps as boxes, queues as triangles, information flow, and a timeline ladder along the bottom separating process time (top) from wait time (bottom). Add rework loops, %C&A (percentage complete and accurate) at each step, and kaizen bursts for improvement ideas. Map the current state first, then a future state with explicit targets, then a plan.
The arithmetic
- Lead time = sum of process time + sum of wait time across the stream.
- Flow efficiency = process time divided by lead time.
- %C&A measures quality at each handoff: if a downstream step receives incomplete input 40% of the time, that is a defect source upstream, not a downstream performance problem.
Improvement engines
PDCA for iterative local improvement; DMAIC where a problem needs statistical definition and control; A3 for structured problem framing and shared understanding; kaizen events for bounded bursts of change. The Theory of Constraints says improve the bottleneck - improving a non-constraint changes nothing in throughput.
Worked example
Four steps: intake (1 hour of work, 2 days waiting), build (6 hours, 3 days waiting), test (8 hours, 3 days waiting), deploy (1 hour, 1 day waiting). Process time totals 16 hours; elapsed time is 9 days, or 216 hours; flow efficiency is 16 / 216, about 7.4%. Reasoning: buying faster build tooling might save three of the six hours, lifting efficiency to roughly 8.8%, while cutting each wait by half would more than double it. The queues dominate, so the future state should target wait time, batch size and handoffs - with build speed addressed later, if at all.
Traps
Mapping the org chart rather than the work; ignoring rework loops and review cycles; mixing calendar hours and business hours when computing efficiency; treating the map as a one-off document; proposing a tool purchase when the data shows a queue.