One warehouse looked efficient until service failures and interstate line-haul were counted together. Opening a second node looked expensive in isolation. The cheap option on the lease was not the cheap option in the network.
The difficulty
A single DC kept overhead simple. National customers on the far side of the network paid for that simplicity in transit time and failed appointments. Finance had a warehouse paper. Transport had a lane paper. Neither paper contained the other. The board was being asked to choose between two incomplete stories.
What we changed
We compared stay-put, a 3PL overflow node and a two-DC design against customer geography, inbound ocean and inventory. Transport, warehouse and inventory were modelled as one number. Service failures were priced, not narrated. The recommendation was a go or no-go the board could fund, with a clear view of what they would give up if they stayed.
Difficulty
One node. Split business cases. Service treated as a complaint, not a cost.
Measure
Stay-put versus overflow versus two-node, with inventory in the same model.
Result
A decision paper. Implementation, if they proceed, remains theirs to fund.
The result
The board received a number that did not belong to a landlord. Whether they opened the second node is not published. The method is: do not let rent win an argument that service and line-haul have not been invited to.
What the split papers were hiding
The warehouse paper showed a cheap, full node. The transport paper showed interstate line-haul and failed appointments as a carrier problem. Inventory sat in a third conversation about buffers that could not be picked fast enough for the far-side customers. None of the papers contained the others. The cheap option on rent was expensive once failed windows and extra line-haul were priced as part of the same number. The expensive option on rent was cheaper once those costs moved.
We modelled stay-put with operating changes, a 3PL overflow for true peak, and a two-node design with inventory policy included. Service failures were given a cost, not a complaint narrative. Inbound ocean into the existing node was left in the model so that a second DC would not be justified while containers still dumped into the first yard. That last point is where network papers usually cheat.
What the board could fund
The decision paper named what they would buy and what they would give up under each option. Implementation remained theirs: lease, fit-out, WMS, labour, customer notices. We wrote the demand pack path if the extra node was to be a 3PL. We wrote the stay-put operating list if it was not. We did not tour a shed we own. We do not own one. The output was a number the landlord did not write.