Re: In training, you learn for life
If... I could actually select "my parcel non-delivery company can't be bothered" and the parcel gets delivered to a nearby collection center ... then everybody is happy.
This is the Amazon locker. There's an assumption in the use of these that Amazon will succeed in delivering to a locker rather than a house - after all they know exactly where they are. This is a false assumption.
On one level some products, sometimes seemingly more or less at random, are banned from lockers but instead of telling you this every locker is reported as "full" when an attempt is made to select it.
On another, even when the locker is selected Amazon may fail to deliver. At this point Amazon's propensity to only code for the "happy path" comes into play. The courier is apparently allowed to move on from the locker without having delivered all the packages. How? Does he have to provide some feedback to the system so the customer can be informed PDQ? That would require the situation to be properly handled.
Reality - non-filled locker in Yorkshire, tracking subsequently locates package in France and next day a courier turns up to collect the return of what wasn't delivered (this can also happen to a non-delivery to the door). Clearly there's no proper handling ot this situation, just more or less random stuff.
I'm sure every developer here knows that a large part of the code of successful system consists of catching and handling things that don't go as intended, if only to log things for later consideration. Amazon apparently doesn't.
Anyone dealing with customer service should learn early that when things go wrong you must keep the customer fully informed. Amazon doesn't and if they haven't collected the information they can't.
Anyone dealing with quality knows that what goes wrong should be reviewed and the knowledge gained fed back into process improvement. You can't do that without collecting data.
I find it amazing that Amazon's algorithms for predicting delivery times have improved to the point where delivery by their own transport is almost invariably within a quarter of an hour of the centre of the quoted range although the range might vary during the day as data gets fed back. They clearly have some very able developers working for them. Why then, can they not handle failures sensibly?