A cheap shed in the wrong place is expensive. So is a national footprint assembled from leftover leases. Network design in this practice is a service, inventory and inbound-flow question. It is not a real-estate brochure and it is not a warehouse we happen to have vacant. We do not own sheds. We will not talk you into a tenancy so that a related company can invoice you.
What the model has to hold together
Customer geography. Supplier and port geography. Current DC locations and their real capacity, not the brochure capacity. Peak. Returns. The cost of a missed delivery window. Inbound ocean that the warehouse has to absorb on the day it lands. Interstate line-haul that looks cheap until you count failed appointments. Inventory that is called a buffer and cannot be picked because the system does not know it is there.
We compare stay-put, a 3PL overflow node, and a two-node design. Sometimes stay-put is the answer, with a different slot discipline and a different inventory policy. Sometimes a second node is the answer and the rent looks frightening until it is placed next to the line-haul you stop buying. Three separate business cases, one for warehouse, one for transport, one for stock, are how companies fund the wrong node with great confidence.
Service is a cost
Distributors sell appointments. Manufacturers sell reliability into a customer's production slot. A model that optimises only rent will steal from those promises and call it efficiency. We put failed appointments and late deliveries into the same page as warehouse cost. If that makes the cheap option look less cheap, that is the point of the file.
What you receive
A go or no-go on the node question, with the service you would buy and the cost you would take on. Implementation, if you proceed, is yours to fund. We will write the demand pack if the node is to be a 3PL rather than an own-operation. We will not project-manage a fit-out unless that is separately scoped, and we will not fill the shed with our own labour.
Open this file when one DC looks efficient until the far side of the network complains, or when someone has already booked a warehouse tour and you want a number that is not the landlord's. Do not open it if you want us to quote you pallet positions we own. We do not own any.
Stay-put is sometimes the answer
A second node is not a moral improvement. It is a cost and a service choice. We have recommended stay-put when the service failures were slot discipline, inventory accuracy or a carrier that could not hit the appointment, not a missing geography. Opening a shed to hide a process problem gives you two process problems and a lease. The paper has to be willing to say that. A practice that is paid to fill space will not say that. We are not paid to fill space.
Overflow through a 3PL is a third option that boards under-use because it looks temporary. Temporary is sometimes the honest word. Peak that lasts ten weeks does not always justify a second lease. Peak that has lasted three years is not peak. It is the network. We distinguish those in the model rather than in a workshop adjective.
Inventory belongs in the same model
A node without a view of what stock would sit there is a real-estate diagram. Safety stock that cannot be picked is not safety. Transfer stock between two DCs that the system cannot see is a new freight invoice. We put inventory policy next to warehouse cost and line-haul. If that makes the cheap rent look expensive, the model is doing its job. If finance wants a stock number and operations wants a delivery promise on two different slides, we will put them on one page and wait until someone owns the conflict.
Implementation, if you proceed, is yours to fund. Fit-out, WMS, labour and the customer communication sit with you. We will write the demand pack if the node is a 3PL. We will write the stay-put operating changes if the node is not. We will not project-manage a construction program unless that is a separate brief, and we will not staff the dock.