Numbers measured from what happened, not from the plan.

Activity is not health

Ask most operations how the month went and the answer is a total: we did 2,400 deliveries. That is a measure of how busy you were. A month with 2,400 deliveries, a 71% on-time rate and two drivers who quit is not a good month.

The numbers worth watching are the ones that change a decision — on-time rate, stops per driver-hour, first-attempt success, cost per vehicle, service time at the door — and each of them here comes from work that actually happened.

Why measured beats estimated

Reports built on planned data flatter you in a specific way: they measure the plan you wrote rather than the day you had. Planned distance is always shorter. Planned on-time is always higher. Planned cost is always lower.

Because arrivals are geofenced, completions are timestamped, and distance is taken from the route actually driven, the reports here describe the operation instead of the intention. The gap between planned and actual is itself a useful signal — a large one usually means drivers are not running the plan, and they generally have a good reason that belongs in the plan.

The splits are where the answers are

An aggregate on-time rate tells you whether you have a problem. It never tells you what to do. The same number split by route, driver, window type and weekday usually resolves to something specific and fixable: Thursday afternoons on the north route, or the two-hour windows that were never realistic.

Same for first-attempt success, which splits by failure reason and almost always points at notification timing or a handful of bad addresses rather than at driver behaviour.

Pick one number a quarter

An operation that knows its numbers and is deliberately working on one is in better shape than one tracking twenty and acting on none. Measure everything, act on the metric with the clearest link to money right now — usually first-attempt rate or stops per driver-hour — and let the rest tell you whether it worked.