Warehouse KPIs: measures that change behaviour, for better and worse

Measure each stage, then measure the promise

The stages have natural indicators: elapsed time from a vehicle arriving to stock being available, retrieval rate per hour worked, quantity and identity accuracy at the point of packing, loads leaving before their cut-off. Above them sits the measure customers actually feel, which is whether complete orders left on time. Stage measures explain movement in the top-level one, so reporting the promise without the stages leaves a manager knowing there is a problem and not where it lives.

Every measure has a defence mechanism

Retrieval rates rise when easy work is chosen first and awkward orders are left for the next shift. Elapsed time from arrival to availability improves when goods are booked in before they are checked. On-time despatch improves when the cut-off is quietly moved. None of these are dishonest by intention; they are what a reasonable person does when judged on one number. Pairing each measure with its natural opponent, rate with accuracy, speed with completeness, is what removes the incentive.

Quality figures need a denominator people believe

Accuracy expressed per order flatters an operation shipping large orders and punishes one shipping single items, so the unit has to match the work. It also matters where the error is detected: mistakes caught at packing are cheap, mistakes caught by the customer are not, and reporting only the customer's discoveries hides the internal ones that predict them. Recording where each error was found turns the measure into a map of which controls are working.

Cost per unit conceals the mix

A cost per order or per case moves when the shape of the work changes, not only when performance does, so a site absorbing more single-line orders will appear to worsen while doing nothing differently. Comparing periods therefore requires the mix to be stated alongside the cost. Between sites, the comparison is rarely meaningful at all unless the buildings, ranges and service commitments genuinely resemble one another.

A short board, reviewed where the work is

Measures used by supervisors during the shift need to be few, current and actionable: what is behind, what is blocked, what leaves next. Monthly analysis serves a different audience and can be richer. Keeping those two purposes separate stops the operational board from filling with figures nobody can act on before the vehicle leaves, which is the usual reason such boards stop being read at all.

Frequently asked questions

How many measures should a warehouse report?

Few enough that the management team can recall them and act on each. A short set covering inbound speed, retrieval rate, accuracy, on-time despatch and stock accuracy at location covers most decisions; anything beyond that belongs in analysis rather than on a board.

Can two sites be compared on productivity?

Only with care. Layout, product mix, order profile, storage medium and service commitments all move the figure independently of how well either site is run, so the honest comparison is each site against its own trend and its own measured standards.