JJ Signal · Research · 7 min read
What an exception actually costs to clear
Thirty-eight minutes per exception, of which twenty-nine were gathering context rather than deciding anything.
Published Logistics & Supply Chain
Freight operators staff the exception desk to the volume of exceptions, when the binding cost is the context-gathering each one requires. Three quarters of the handling time is spent assembling a picture across systems, which means the desk is a data problem being solved with headcount.
The evidence
Drawn from our own delivery work. Each claim states its sample and its method, including where the sample is small.
- 01Median 38 minutes to clear an exception, with about 29 spent gathering rather than deciding.One operator, roughly 1,200 consignments a week, desk instrumented for a month. A single operator, but the ratio held across all four desk staff.
- 02Queues worked in arrival order, so what aged was uncorrelated with what it cost to leave.Exceptions older than two days usually became credit notes. Ranking by value at risk against the SLA clock changes which ones age, not how many.
- 03Margin by lane could not be produced inside a quarter at a second operator, for the same underlying reason.Four cost sources with four notions of a job. Reconciling identity took most of a two-week engagement and produced the whole answer.
What follows from it
Before adding a person to the exception desk, measure what fraction of handling time is gathering. If it is most of it, you have a data aggregation problem and another pair of hands will open another four tabs.
What we would do
- 01Instrument the desk for a month. Split every handling period into gathering and deciding.
- 02Aggregate consignment, carrier, driver and SLA state into one exception record before changing any process.
- 03Rank by value at risk against the SLA clock, so what ages is genuinely what was cheapest to leave.
- 04Write back to the TMS and the customer in one action, so the desk stops re-keying its own outcome.
Get JJ Signal by email
Roughly monthly, and only when there is something measured to report. If a month produces nothing worth your attention, you get nothing.
What you are signing up to