You opened a second distribution center to get product to customers faster. It worked. Transit times dropped, and the West Coast stopped waiting five days for everything.
It also added a decision to every purchase order you write, and that decision has no owner.
Before, a purchase order had one destination. Every unit went to the same building. The only question was how many. Now every line on every order carries a second question: how do these units divide across the network? That question gets answered dozens of times a season, usually in under five seconds, usually by a number somebody typed into a split profile two years ago.
What a Fixed Percentage Split Assumes
Most teams answer it with a fixed percentage. 60%/40% to the East/West facility. It is fast, it is stable, and it made sense the day it was set.
Look at what that number quietly assumes:
- The stores each warehouse serves still sell in the same proportion they did when you set it.
- You have not opened, closed, or reassigned doors since.
- Both warehouses are starting from a similar inventory position.
- Nothing already on the water changes the picture.
- This product sells the same geographic mix as the average of your assortment.
Every one of those drifts. You opened eleven doors in the Southeast last year. A regional wholesale account came online. Three stores moved from one fulfillment center to another and nobody updated the split. The percentage stayed sixty forty through all of it.
The drift is invisible at the order. Nothing errors. The PO splits, it gets approved, it ships. You find out months later, in the form of a warehouse sitting on cover it cannot move and another one shorting stores on a style that is selling.
Then you pay twice. The correction is a transfer. You pay freight to move units between buildings that you already paid freight to fill, plus the handling on both ends, plus the two or three weeks the product spends in transit instead of on a floor. A bad split is not a rounding error. It is a freight bill and a markdown you have not seen yet.
And the fix does not stick. Teams that notice the drift re-derive the percentages by hand, usually once a year, in a spreadsheet nobody else can follow. It is correct for a quarter. Then the fleet moves again.
Splitting a Buy Across Distribution Centers Without Guessing
The discipline here is doing the arithmetic the percentage was standing in for. You can run all of this in Excel on your next order.
Map Stores to Fulfillment Center, and Keep the Map Current
This is the unglamorous foundation and the thing most teams do not actually maintain. Every store belongs to the warehouse that ships it. When a door changes warehouse, the map changes that week, not at annual review.
Attribute Planned Demand Instead of History
Roll each store's planned sales for the order's coverage window up to its DC. That gives you demand by warehouse for this product, over the window the order actually covers. History tells you where you shipped last year. The plan tells you where the units are supposed to sell.
Project Each DC Forward to the Arrival date Instead of Today
Current on-hand is the wrong number. What matters is what each warehouse will hold the day the order lands: on hand today, plus everything arriving on open POs before then, minus what its stores are planned to sell in the meantime.
Buy the Gap
Here is where the percentage approach fails hardest, and it is the whole point.
Split on the Shortfall Instead of Demand Share
Two DCs with identical demand do not need identical units. If one is already covered through the window by stock on hand and inbound POs, its share of this order is not half. It is whatever is left after you subtract what it already has coming. Demand tells you the target. Position tells you the order.
Re-Derive When the Order Changes
Change the units, move the delivery date, and the arithmetic changes with it. A split computed against the old quantity is a stale split. This is the step that kills manual methods, because nobody wants to rebuild the spreadsheet because a date moved a week.
Run this once and it is an afternoon of satisfying work. Run it on every line of every order, and re-run it each time a quantity or a date moves, and it stops being a discipline and starts being a job nobody has. That is the point where teams go back to the percentage.
Toolio derives the split from your planned demand and each DC’s projected position at the order's start date, recalculating as you edit and flagging the lines it could not compute, so the split stays current on orders you write in seconds rather than degrading between annual cleanups.
