Two hundred exceptions and one morning: how to order the list

Why exception counts are the wrong headline, how to rank PO exceptions by lateness and value, and the four priority bands that make a list into a work queue.

Automating the checking is the easy half. A daily run that turns ten thousand PO lines into two hundred flagged ones is a genuine improvement, and it is also where most projects stop — leaving a buyer with a shorter spreadsheet and the same question. Which of these first?

Counts are the wrong headline

"215 open exceptions" tells you nothing actionable. The number moves for reasons nobody controls — a big order release, a seasonal peak — and it treats a pending confirmation on a low-value consumable as equivalent to a late delivery of a critical part.

Two questions make the list useful: how late is it, and how much money is behind it.

Rank by lateness and value

Lateness is not a binary. A line whose delivery date has passed is in a different category from one due next week, which is in turn different from one whose confirmation is simply overdue. Value is the multiplier: the same delay on a ten thousand rupee line and a ten lakh line are not the same event.

Four bands, not ten. The purpose is that a buyer can work down from the top and stop when the day ends, knowing that nothing expensive was left behind — and ten bands do not survive a busy morning.

Say what is fine, too

The band that gets left out of most designs is the last one. A run that reports only problems leaves an open question about the other nine thousand eight hundred lines, and someone will spot-check them by hand to be sure.

Stating that the rest are on track, and how that was determined, is what lets people stop looking. That is most of the time saving.

Attach the money

Pending value per exception type changes the conversation, because counts get argued with and rupees do not. Roll it up two ways — by vendor, so you know who is holding up the most, and by plant, so you know where it lands.

pending value   = open qty x net price, summed per band
by vendor       = who to call first
by plant        = who to warn first
ageing overlay  = 0-3 / 4-7 / 8-15 / 15+ days pending

The ageing overlay is what stops the same lines being re-triaged forever. Anything sitting in the 15+ bucket has survived several rounds of good intentions, and that is a process problem rather than a supplier problem.

One paragraph beats one dashboard

A dashboard is the right tool for digging. It is the wrong tool for a Monday morning, because it asks the reader to work out what changed. A short written summary — how many lines were checked, how the exceptions split, and which supplier accounts for the largest share — is what most people need on most days.

The dashboard should still be there. It just should not be the thing anyone has to open to know whether today is normal.