Cluster before you route
Most route tools start with miles. I start with clusters.
If you route before you group the work, you optimise a mess. Packaging plants ship mixed SKUs to many sites in short windows. The useful first move is to cluster what belongs together on the same run. Then you route the cluster. Not the other way round.
What clustering actually means on a plant
Clustering is not a fancy name for sorting by postcode. It is deciding which drops can share a trailer given windows, pack sizes, axle limits, and last-on first-off. Two sites on the same corridor that cannot share a deck still should not sit in the same cluster. Two sites a bit further apart that fill an axle-legal trailer cleanly often should.
I see planners jump straight into a mileage map because the map looks decisive. The map does not know crush rules, stack heights, or that the dense board has to sit over the drive axles. Those facts decide the cluster before any turn-by-turn path matters.
A good cluster also respects the morning reality. What is actually ready. What will finish in time. What customer window is hard. Route distance on unfinished work is theatre.
Corrugated and converting mixes make this sharper. Light converted packs and heavy board do not behave the same on a multi-drop. If you cluster only by geography, you inherit a fill problem before the truck moves.
Why shortest path comes second
Once the cluster is honest, routing can cut wasted miles inside that set. That is useful. It is not the main lever. Empty bay space, bad drop order, and axle breaches waste more than a slightly longer lane between two sensible stops.
If you reverse the order, you get pretty routes that still leave under-filled trailers and restacks at the first drop. The yard feels busy. The plan is still weak.
Planners know this in their bones. The software often pushes them the other way because miles are easy to score and clusters are harder to explain in a board pack. On the dock, the hard thing is the right thing.
Where Clockwork sits
Keep the ERP. Clockwork sits after it. Order and ship data move once via Fabric. Then you cluster, fill, and build multi-drop with axle-legal 3D before you obsess over the last kilometre.
Cluster before you route. That sequence matches how a packaging dock actually works.
