Monitoring, Measurement, and Control: Metrics That Change Behavior
What you must know
This domain tests metric design and interpretation. Know the standard families: perfect order (on-time, complete, damage-free, correct documentation), order cycle time, inventory turns and days of supply, fill rate versus service level, cost-to-serve, and carrier performance. Understand leading versus lagging indicators, and why a balanced set beats a single hero metric. Know the control loop: set target, measure, compare to threshold, diagnose, act, re-measure. Be comfortable with variance analysis and basic statistical process thinking - distinguishing common-cause variation from special-cause signals.
Traps
The classic trap is a metric that is gamed: measuring warehouse productivity by units per hour invites larger pack sizes and worse accuracy. Another is comparing fill rate definitions - line fill, unit fill, and order fill are not interchangeable. Watch for dashboards that report averages when the decision needs the tail (95th percentile lead time). Also expect confusion between a target and a forecast, and between measurement and control: a report that nobody acts on is not control.
Worked example
On-time delivery is 96% overall, but the top 10 customers are at 99% and the rest at 91%. The aggregate hides a segment problem. Segment the metric, find that one lane accounts for most failures, and set a lane-level threshold. Rule of thumb: segment before you average, and pair every efficiency metric with a service metric so improvement in one cannot silently damage the other.