Skip to content
allanninal.dev/build
Collaborate

Part 5 of 7 · Delivery exception handler series ~5 min read

What the exceptions reveal

Handled individually, delivery exceptions are a cost. Counted properly, they are the clearest available picture of where a fulfilment operation is actually failing.

Key takeaways

  • Separate internal causes from carrier causes before reporting anything.
  • Address quality is usually the largest single fixable cause.
  • Exceptions cluster by depot and postcode more sharply than by carrier.
  • Report rate per thousand shipments, not counts.
  • The never-collected count is a warehouse metric wearing a carrier’s uniform.

Whose failure was it

Three sources of delivery exceptions and their approximate sharesThree boxes stacked on the left. Ours, before despatch, covering never collected and wrong item, labelled about a quarter. The customer's data, covering address, phone and access notes, labelled about a third. And The carrier's, covering mis-sorts, damage and missed rounds, labelled the rest. All three converge on One exception report split by cause, and that leads down to Two thirds are ours, to fix rather than to negotiate. A note says reported together they all look like carrier problems, and they are mostly not.Ours, before despatchnever collected,wrong itemabout a quarterThe customer's dataaddress, phone,access notesabout a thirdThe carrier'smis-sort, damage,missed roundthe restOne exception reportsplit by causeTwo thirds are oursto fix, not to negotiateReported together, they all look like carrier problems. They are mostly not.
Fig 1. The three sources of delivery exceptions. Splitting them is the difference between a carrier negotiation and a fixable internal problem.
  • Compute
  • Machine learning
  • Analytics
  • Front-end & mobile

Address quality

The largest single fixable category in most businesses, and it is fixable at the point of order rather than at the point of delivery. Flats without a number, business addresses with no company name, postcodes that do not match the street, and phone numbers that were never asked for.

The exception data points precisely at which of those matter. If failed deliveries cluster on addresses with no second line, adding a prompt at checkout for flat and unit numbers is a half-day of work that removes a recurring cost permanently.

Depots, not carriers

Delivery exception rates for one carrier broken down by depotA bar chart with four bars showing exceptions per thousand shipments. The carrier overall is three point one. The north depot is two point four. The central depot is two point eight. One particular site is eleven point six. A note says the carrier's overall rate is fine and one depot is four times everything else.05101520~3.1Carrier overall~2.4Depot: north~2.8Depot: central~11.6Depot: one siteExceptions per 1,000 shipmentsThe carrier's overall rate is fine. One depot is four times everything else.
Fig 2. The same carrier broken down by delivery depot. An acceptable overall rate can hide a single site that is producing most of the problems.

This finding is actionable in a way a carrier-level number is not. Carriers know which of their sites are struggling, and a conversation that names one depot with a rate and a sample size gets a specific response, where “your service has been poor” gets a general one.

In some cases the answer is to route deliveries for those postcodes through a different service, which is a decision that only becomes visible once the data is cut this way.

Rates, not counts

As with damage claims, counts blame whoever you use most and whichever region you sell into most. Exceptions per thousand shipments, computed per carrier, per service and per depot, is the only version of this report that supports a decision.

What the report says

The monthly page

  • Shipments: 4,180. Exceptions: 61, a rate of 14.6 per thousand.
  • Internal, before despatch: 16 — 11 never collected, 5 wrong item.
  • Address or contact data: 22 — 14 of them flats with no unit number.
  • Carrier: 23, of which 14 were one depot.
  • Resolved without customer contact: 38. Contacted first: 19. Customer contacted us first: 4.
  • Average days from exception to resolution: 2.8, against 6.1 before this existed.

The third line is the one that produces a change. Fourteen failed deliveries in a month to flats with no unit number is a checkout form problem with a number attached, and it is not something anybody would have guessed from handling the exceptions one at a time.

The last line is the honest measure of whether the system is working. Not how many exceptions there are — that is mostly outside your control — but how long they stay unresolved, which is entirely within it.

The uncomfortable finding

Most businesses running this analysis for the first time find that a substantial minority of what they had been treating as carrier failures happened in their own warehouse. That is unwelcome and it is the most valuable thing in the report, because internal problems can be fixed without anybody’s cooperation.

Next: what all of this costs to run.

All posts