Choosing a Metric Set for a Distribution Network Under Service Pressure
A consumer-goods maker runs three DCs serving 1,200 retail ship-to points. Leadership is fixated on a single headline: fill rate has slipped from 96% to 91% over two quarters. Operations responds by adding safety stock at all three DCs, which lifts fill rate to 94% but pushes inventory days from 38 to 57 and warehouse overtime up 22%. Finance flags working capital. Meanwhile, the customer service team reports that 60% of the missed lines were single-SKU orders that could have shipped from an alternate DC, and that order-cycle-time variability (not average lead time) drives most retailer complaints. The company has no agreed definition of fill rate: sales counts units, logistics counts order lines, and planning counts complete orders.
1. Diagnose why the fill-rate obsession produced a costly, partially effective response, and recommend how the metric set should be restructured.
Read the worked answer
The core problem is a single lagging metric with three incompatible definitions, so each function optimizes a different number and nobody can localize the loss. The 5-point fill-rate gain cost 19 inventory days and 22% overtime - a poor trade because the root causes were network allocation and variability, not insufficient stock everywhere.
Restructure in three moves:
- Fix definitions first. Agree one primary service metric (e.g., complete-order fill rate) with a documented unit/line/order hierarchy, so sales, logistics, and planning read the same number.
- Add balancing metrics. Pair service with inventory days and a variability measure (order-cycle-time spread, e.g., standard deviation or on-time-in-full). A metric set that can only be improved by adding stock is a broken set.
- Segment the loss. The 60% alternate-DC-eligible misses point to a network/ATP logic gap, addressable by multi-source allocation rules at far lower cost than blanket safety stock. Variability-driven complaints point to process reliability, not buffer.
Then re-evaluate the added safety stock: keep buffer only where the loss is genuinely demand-driven, and release the rest.
2. The CFO asks whether the company should benchmark fill rate against industry peers before acting. What is your advice?
Read the worked answer
Benchmarking is useful for setting ambition, not for diagnosis. A peer fill-rate percentile tells you whether 91% is competitive, but it cannot tell you which of your 1,200 ship-to points are losing lines, why, or what the fix costs. Acting on a benchmark gap alone tends to produce exactly the blanket-buffer response already seen.
Advice: use external benchmarks to frame the target range and to sanity-check whether the inventory/service trade-off is out of line with the industry, but drive the improvement plan from internal segmentation - loss by DC, by SKU class, by order type, by cause. Also ensure any comparison uses a like-for-like definition; comparing your order-line fill rate to a peer's unit fill rate is meaningless. Sequence: internal diagnostic and definition alignment first, benchmark as a calibration input, not a decision trigger.