Supply chain KPIs: a measurement set that survives scrutiny

Balance across four competing dimensions

Service, cost, working capital and quality pull against one another, and a set that reports only one invites improvement paid for out of the others. A minimal balanced set covers whether customers got what they asked for when they asked for it, what the flow cost to run, how much capital is committed to stock, and how much of the work had to be done twice. Reporting these together makes trade-offs visible instead of allowing them to be hidden in another function's budget.

Definition is where measures are won or lost

Delivery performance measured against the customer's requested date and against the date the business confirmed can differ enormously, and a business quoting one while its customers experience the other is measuring its own convenience. Similar traps exist in whether partial deliveries count, whether cancelled orders are excluded, and where the clock stops. Every measure needs a written definition covering source, inclusion rules, timing and the level at which it is calculated.

Measure what a decision can change

A figure that nobody can influence produces resignation rather than action. The test for each measure is to name the decision it informs and the person who takes it. Where the answer is unclear, the measure belongs in analysis rather than on a scorecard. This test usually removes half of an existing set, which improves attention on the remainder more than any presentational change could.

Anticipate the gaming

Every measure creates an incentive, and it is worth asking in advance how each could be satisfied without improving anything. Availability can be protected by overstocking, stock reduction by starving the business, delivery performance by quoting longer lead times, and cost per unit by pushing volume through at the expense of service. Pairing each measure with its natural counterweight is the standard defence, and it works better than exhortation.

Leading indicators and the review rhythm

Outcome measures report what already happened, so a set built only from them arrives too late to change anything. Adding a small number of forward indicators — schedule adherence, supplier confirmation reliability, the ageing of open exceptions, parameter freshness — gives warning while there is still time to act. Reviewing outcomes at the planning cadence and forward indicators more frequently matches each measure to the horizon on which it can be influenced.

Frequently asked questions

What is the single most useful customer-facing measure?

A composite that requires the whole order to arrive complete, on the date promised, undamaged and correctly documented. Measuring the elements separately allows each to look acceptable while the proportion of orders that were faultless end to end is much lower.

How many measures should a scorecard contain?

Few enough that a management team can hold them in mind and discuss each one properly in a review. Additional detail belongs in the layers beneath, available when a headline measure moves and someone needs to find out why.

Should targets be absolute or relative to a baseline?

Improvement against your own baseline is generally more actionable, because achievable levels depend on product mix, network shape and service model. External comparison is useful for direction, but a target imported from a business with a different structure tends to be either trivial or unreachable.