We had a client, a home goods brand out of Austin, who called me in October 2024, two weeks before their holiday peak, and asked a question I hear at least once a quarter: why did your system send this order from the wrong warehouse? A customer in Atlanta had ordered a lamp and curtains. All three warehouses had both items. OMS order sourcing picked the facility in Reno. The freight bill came in higher than shipping from Dallas, and delivery landed a day later than expected. The client was not angry about one order. He was angry because nobody could explain the decision in plain English.
I pulled the sourcing log before I called him back. The system had done exactly what it was told to do. Dallas was running low on that lamp, with a big wholesale replenishment still in transit, so the inventory balancing rule steered the order west to protect Dallas stock for the orders already queued behind it. The logic was defensible. The bill was not. That gap, between a rule that makes sense in isolation and a result a client will actually pay for, is where most multi-warehouse allocation arguments live. All figures here are illustrative examples, not real quoted rates.
Why OMS order sourcing rules fight each other
An OMS usually juggles four sourcing rules, and each one sounds reasonable until it meets the other three. Ship from the closest warehouse for the fastest transit and easiest return. Ship from the cheapest warehouse, counting freight, pick and pack, and packaging together, not just the carrier rate. Ship to protect the delivery promise, paying extra so a two-day commitment stays a two-day delivery. Balance inventory across the network, so one building does not run dry while another sits on months of stock.
Put those four rules on the same order and they will often point at four different buildings. The closest warehouse may have the highest labor cost, the cheapest option may add a day in transit, and the warehouse with the deepest stock may be the farthest away. A system that follows one rule to the letter will quietly sacrifice the other three, and the client only finds out when the invoice arrives.
I learned that the hard way with a beauty brand in the spring of 2025. Their hero serum was stocked in California, Texas, and New Jersey, and the default rule was proximity. Orders from Los Angeles flooded the California building, which sold through its serum in four days. The orders that followed had to ship from Texas, adding two transit days and breaking the 48-hour delivery promise the brand had just made in a livestream. The contract said "ship from the nearest warehouse" and nothing about when inventory balancing should override proximity, what the low-stock trigger was, or who was supposed to watch it. A rule without a trigger and an owner is not a rule. It is a hope.
My view is blunt: no single rule should run the network. OMS order sourcing needs a ranked list, with a tie-breaker written down, and a review date attached to it. If the ranking lives only in a system setting nobody has opened in a year, the client will eventually pay for that neglect.
Splitting an order can cost less, and cost more
The expensive decision is rarely which warehouse gets the order. It is whether to split the order at all. One order, three items, three warehouses: the system can split it into three parcels or wait and consolidate it in one building. Splitting is faster for the first item. Consolidating is cheaper on freight and simpler for the customer, who gets one box and one tracking number.
Here is the counterintuitive part: splitting sometimes saves money. In December 2024, a pet supplies client had an order with a 15-pound bag of dog food and two small bags of treats. Only Dallas had the dog food. Consolidating in Dallas sounded efficient, until we priced it: that lane carried a remote-area surcharge and a heavy-item fee that wiped out the savings. Splitting meant the dog food traveled by ground from Dallas while the treats shipped as a small parcel from the nearest building. The two shipments together cost less than the single consolidated shipment, and the customer received the treats a day earlier.
That does not mean you should split everything. Three parcels mean three tracking numbers, three times the customer service calls when a shopper thinks two items are lost, and pick fees charged per shipment. I draw the line in three places. First, the freight saved by splitting has to cover the extra pick fees and service time, judged on total cost, not the carrier rate alone. Second, any order with a heavy item gets priced by weight and zone before anyone talks about splitting. Third, orders that are one gift, or a matched set, stay consolidated, even if that costs a little more.
Those boundaries belong in the sourcing configuration, not in one operator's head.
When sourcing goes wrong, who pays?
This is my day job, and it is where clients lose the most money without noticing. A sourcing mistake falls into three buckets. If the system followed the agreed rules and the result was still expensive or slow, the rules were wrong, and whoever approved them owns the problem. If the system ignored the rules, for example shipping on cost when the contract said delivery promise comes first, that is an execution failure, and the operator owns it. If the inventory data was wrong, liability turns on what the contract says about inventory accuracy.
I have signed well over a hundred warehouse contracts, and the weakest clause I keep seeing is a single sentence: "Provider will source orders per system rules." That sentence says nothing. Which rules, who set them, how often they are reviewed, who pays the freight difference on a mis-sourced order, and how a late delivery is compensated, all of it is missing.
The Atlanta lamp order ended in a split: we covered half the freight difference, the client covered the other half, and we both agreed the real failure was a balancing threshold set for slow-season stock levels and never revisited before peak. After that, the client added a clause: sourcing rules get reviewed every quarter, and again before every peak season, mis-sourced freight is charged to the responsible party, and late-delivery compensation is fixed in an appendix.
| What went wrong | First question to ask | Who usually bears the cost |
|---|---|---|
| System followed the rules, result was costly | Who approved the rules, and when were they last reviewed? | The party that set the rules |
| System ignored the rules | Do the system settings match the contract? | The operator running the system |
| Inventory data was wrong | What does the contract say about inventory accuracy? | Whoever owns the inventory clause |
A configuration checklist for multi-warehouse allocation
A workable sourcing setup needs a few pieces clients rarely ask for until it is too late. Rank the four rules in writing, with a tie-breaker for conflicts, and never accept "the system decides" as an answer. Set a trigger for each rule, such as the stock level below which a warehouse stops receiving orders. Write the split rules as their own section: when splitting is allowed, the maximum number of parcels, and whether the customer has to be told first. Put the failure handling in the same document: who pays extra freight, how late orders are compensated, and how often the rules are reviewed.
Two details worth naming: before a new sourcing rule goes live, run it against last month's real orders and see where it would have sent them and what the freight total would have been. Do not use peak season as your test run. Second, keep a change log for every rule change: who changed it, when, and why, and whether the client was notified. One sentence in the contract covers it.
Sourcing is not really a software problem. It is a commercial problem with a software tool attached. A system can find the cheapest warehouse. It cannot decide how much a customer will pay for one faster day, or how much a stockout costs in lost repeat business. Those numbers have to be agreed by the client and the provider together, written into the contract, and reviewed on a schedule.
Next time you negotiate a contract, ask these three questions: who set the sourcing priority, and how often is it reviewed; where is the split-order boundary written, and how many parcels need my approval first; and when sourcing goes wrong, who pays the extra freight and late-delivery compensation. Get clear answers, and the next time the rules fight each other, you will have something to point at.
By Susan, 3PL Sales Director







