A regional delivery operation is late too often. Here eight rooms take up that one problem, each in a different format, so each looks at a different side of it: what causes what, what to aim for, what to do first, in what order, whether an argument holds, what the evidence shows, which system to choose, and what matters most to customers.
The method does not change. Fifteen people from the operation write one idea per card, then judge the cards a pair at a time, and a link is kept when 12 of the 15 agree, three quarters. What changes is what a card is and what each pair asks. Each example is cut to a handful of cards.
What is driving delivery delays across the region?
Does “loads are dispatched before the paperwork is checked” significantly influence “drivers get routes they cannot finish”? 12 of 15 said yes: kept.
Root causes: Loads are dispatched before the paperwork is checked · Customs forms are filled by the wrong desk. What an action must move. The symptom everyone ranked first sits above both. The full example has all seven stages and the whole map.
What would deliveries that arrive when promised look like, a year from now?
Would “every load leaves with its paperwork checked” help achieve “customers get a two-hour window they can trust”? 13 of 15 said yes: kept.
Foundations: A shift handover never loses an open problem · Every route can be finished in one shift. What the rest of the future rests on. A plan aimed at the two-hour window alone would miss both.
What should we do about the delays, and what first?
Would “hire a night supervisor” make “add a paperwork check at the loading bay” easier to do? 14 of 15 said yes: kept.
Actions to take first: Hire a night supervisor for the depot · Stop the scanner app logging out every hour. Which actions make the others easier, a different question from which matters most. The paperwork check is the one most people name first; the map says to start with the person who makes it possible.
How do we move every route to planning by stops instead of distance?
Must “train the dispatchers” be done before “try the new routes on two vans”? 3 of 15 said yes, 12 no: no such order.
First steps: Record a month of stop-by-stop times. Where to start. A no is as useful as a yes: the training and the two-van trial can run side by side.
Should dispatch start at 7am instead of 6am?
Does “most late loads left the depot before 6:30” give a reason to accept “a 7am start would cut late deliveries”? 12 of 15 said yes: kept.
Premises: The first loads’ paperwork is checked by one person at the end of a night shift · A 7am start would send the first vans into the school run. What each side stands on. The objection is a claim of its own, with its own reason, and the map does not pick a winner. Here the case for 7am rests on the paperwork, which is a reason to fix the paperwork.
What do we actually know about where the delays start?
Is “early routes lost no more time on the road than later ones” evidence for “on early routes the time is lost at the depot”? 13 of 15 said yes: kept.
Grounding findings: Late deliveries left the depot late in 38% of cases (dispatch log, March to May) · Early routes lost no more time on the road than later ones (van tracking) · 1 in 9 early loads came back with a paperwork error, against 1 in 40 later ones (returns desk). The findings the conclusion stands on, each with its source. If the dispatch log turns out to be wrong, a reader can see what falls with it.
Which route-planning system should the depot choose?
Would meeting “plans by stops, not by distance” help meet “every route can be finished in one shift”? 14 of 15 said yes: kept.
Criteria to judge by: Plans by stops, not by distance · Lets a dispatcher change a route in under a minute · Shows drivers the site notes for each stop. What to judge by, settled before anyone argues for a product. Then three systems are judged against those three, square by square: the current planner meets none, a stop-based planner the first two, a planner with a driver app all three.
Which delivery problems matter most to our customers?
Does “the goods arrive damaged” matter more than “nobody warns the customer”? 9 of 15 said yes, 6 no: neither side reached 12, so it stays on the record as split.
Top priorities: Nobody warns the customer that the delivery will be late · The goods arrive damaged. It ranks; it does not find causes. Customers minded not being warned more than the lateness itself, and the warning fails for reasons only a root cause room would find.
None of these rooms could stand in for another. The root cause room found what causes what, the priorities room what customers feel, the evidence room what is known, the actions room where to start. Rooms of one format on one question fold their maps into one; maps of different formats are kept apart, because their arrows mean different things.
The root cause example · Deliberation formats, drawn · How it works