Why two teams report different on-time delivery

Sales reports 92% on-time delivery; logistics reports 85%. Before looking for an arithmetic error, establish which commitment each team is measuring. Requested and confirmed dates differ. Shipment and customer receipt are different events.

Define the metric in six parts

The following is an illustrative definition, not an industry-wide standard.

If the eligible population is empty, show that the rate is unavailable. Zero percent would imply eligible orders existed and none succeeded.

Averaging the averages changes the question

Suppose plant A meets 9 of 10 commitments and plant B meets 72 of 90. Their rates are 90% and 80%. The simple average is 85%, but the combined order-line rate is 81 of 100, or 81%.

Both calculations have a meaning. State whether plants or order lines receive equal weight. If quantity or value weighting is needed, define another metric explicitly.

Preserve the scope from chart to exception

Selecting this month and plant A in a chart should carry the same conditions into the detail. Users need to inspect included orders, failure reasons and exclusions before acting on the number.

Distinguish query time from source freshness. A page refreshed seconds ago can still show yesterday's closed dataset. Decide how late-arriving actuals affect the reporting period, too.

What to record before designing the dashboard

Keep the definition, source fields, representative inputs, expected results and responsible owner together. Before the monthly close, ask both teams to reproduce the result from the same sample. The visual treatment comes next. An agreed definition makes differences explainable without relying on someone's memory.

Explore aggregation and source reconciliation