Group deviation rows by delivery point and verify each row belongs to the right address, run and container. Recurrence does not prove driver inattention or an address-based cause.
Put the point on the list
At month end, divide deviations at each point by runs that visited it. Two deviations in 10 runs and two in 100 are different rates; report the sample size too.
Observe what happens at the point
- Record the parking location and the carrying path for the vehicle and container.
- Note when the receiver is ready and how long the wait lasts.
- Compare time on the dock with acceptance into the cold room.
- Record arrival time, stop order and earlier openings.
Address or sequence?
If the point is usually last, address and sequence are confounded. Match comparable runs visiting it in different positions; also compare other points in the same position. A remaining difference does not prove the address caused it.
Before sharing with the other party
Check point identity, run and logger record. If timezone, logger clock and download time differ, document the conversion; one matching error can affect the sample.
On a delivery observation form, enter parking location, wait start/end, lid movements and dock-to-acceptance time. If no record exists, do not present recall as a measurement.
Tie a change to a hypothesis
An appointment, parking arrangement or advance notice can be trialled only as a hypothesis supported by site records. Compare similar runs using the same method afterward; equipment decisions follow the quality process.
Date the change. Record the date, changed step and affected runs. Show visit counts and run conditions in a before/after comparison; do not treat a simple difference as a proven effect.
