Here is the conclusion up front: what happens after transferring inventory between warehouses is decided before the truck leaves. A transfer order that says "900 units available" does not mean 900 units can actually leave the donor warehouse. What can move is whatever is left after you subtract what the donor warehouse still needs to sell, and the safety stock it must keep. Skip that subtraction, and the donor's own orders start queueing the day the truck leaves. I have seen this exact failure three times, and the script was nearly identical every time.

The first time was at a home-goods client with two warehouses, one in Los Angeles and one in Dallas. Dallas was short on a popular storage box. Los Angeles showed 1,200 units on hand, the system calculated "900 transferable," and the transfer order went out with one click. Los Angeles stocked out the next day. That storage box was on an online promotion, and Los Angeles was selling twice its normal daily volume. The transferred stock was still on the road while Los Angeles orders piled up for three days. The operations manager asked me on the phone: "Didn't the system say we could transfer it?" The system was not wrong. It just never asked how much Los Angeles needed to sell over the next three days.

The second case was worse. The product was seasonal — space heaters. The donor warehouse transferred most of its stock in October. When temperatures dropped in November, its own orders doubled, and the replenishment lead time was two weeks. A regional manager had approved the transfer, and the WMS had no holdback check at all. The books balanced perfectly. Our earlier guide on multi-warehouse transfer reconciliation and variance handling explains how to reconcile those books — but balanced books do not help when the shelf is empty and customers are angry.

The third time happened on the night before one of my own go-lives. During testing I casually created a transfer order. The system released it, deducted the stock, and the receiving warehouse confirmed receipt. Everything looked fine. Then, running through my go-live-night checklist, I noticed the donor warehouse's on-hand quantity for that item had fallen below its safety stock line, and nothing had stopped it. That night I added a rule: transfer quantity may not exceed available quantity minus safety stock, and anything below that line needs manual approval. Since then, that rule has been a permanent item on my go-live checklist for every multi-warehouse project.

Before Transferring Inventory Between Warehouses, Calculate the Holdback

A transfer decision really comes down to three numbers: how much the donor warehouse has now, how much it will sell before replenishment arrives, and how much it must keep as a floor. The math is simple. The hard part is that most people never do it — they look at the on-hand number and place the order.

Start with "how much it will sell." Take the donor warehouse's average daily sales for that item and multiply by the replenishment lead time in days. Use peak-season averages in peak season and promotion averages during a promotion. Using a slow-season average to plan a peak-season holdback is the same as not planning at all. If an item sells 40 units a day on average and replenishment takes 10 days, expect roughly 400 units of demand before the next delivery — figures in this example are illustrative; use your own warehouse's real sales data.

Then calculate the floor: safety stock. Safety stock is not a number you guess. It covers two things, demand variability and replenishment delay. Items with volatile demand get more. Items from unreliable suppliers get more. Exclusive bestsellers get more. Slow movers get less, or none. My approach is to sort items into three tiers: A-tier bestsellers keep 3–5 days of average daily sales, B-tier regular items keep 2–3 days, and C-tier slow movers keep 1 day or nothing. Write the tier rules into the system configuration, not into someone's head.

Only then do you get the transferable quantity: on-hand stock, minus in-transit allocations, minus orders already allocated but not yet shipped, minus expected demand over the replenishment window, minus safety stock. If the result is negative, the donor warehouse does not even have enough for itself, and the transfer order should never have been created. If the result is positive but small, transfer only that amount. Never move the floor stock too.

Item Figure (illustrative example)
On-hand stock 1,200 units
Allocated, not yet shipped 150 units
Expected demand, next 10 days 400 units
Safety stock (A-tier, 4 days of demand) 160 units
Actually transferable 490 units

I show this table to every client when we talk about transfers. On paper there are 1,200 units, and it looks like plenty. After the deductions, only 490 can actually move. Most people stare at it for a moment, then say, "So we have been transferring our floor stock all along." Yes. That is exactly how most donor warehouses stock out.

After Transferring Inventory Between Warehouses: Three Ways to Recover

The transfer has shipped, the stock is on the road, and the donor warehouse's inventory has dropped below its safety line. What now? Do not panic. There are three moves, in this order.

First, intercept the transfer. If the goods have not left the building, cancel or reduce the transfer order. If they are already in transit, call the carrier and ask whether the shipment can be rerouted or returned. A rerouting fee hurts, but it hurts less than a stockout. I once saw a warehouse call a truck back two hours after it left. The change fee was a few hundred dollars. It saved more than 200 orders the next day.

Second, replenish urgently. Place a rush order with the supplier, or pull a top-up transfer from a third warehouse. Rush replenishment is expensive — air freight and expedite fees never look good on paper. But do the full math: the orders and customer trust lost to a day of stockout usually cost more than the expedite fee. Before asking for approval, calculate how many orders the stockout will affect and what they are worth. A number gets a rush order approved faster than "we are about to stock out."

Third, manage demand. Tell the sales side: limit quantities on this item for a few days, push a substitute product, or move the promotion back. Many operations teams dislike this move because it feels like admitting defeat. But telling a customer up front that an item is temporarily unavailable and recommending a substitute is a far better experience than letting them order and wait a week. Put the quantity limit and the substitute recommendation in the system, not in a chat message, or customer service and the warehouse will be telling two different stories.

If all three moves fail and the donor stocks out anyway, own it, review it, and change the rule. The review asks three questions: did we calculate the holdback before transferring? Were the safety stock tiers right? And who has permission to issue a transfer order without any check? The third question matters most. In many warehouses, transfer permission is wide open, and anyone can empty the floor stock with one order. My recommendation: any transfer that would take the donor below its safety stock line must go through approval, and the approver must be someone who can see both warehouses' inventory — not just the person staring at the shortage in their own warehouse.

Here is the counterintuitive part: the most dangerous moment for a transfer is not when the receiving warehouse is out of stock. It is when the donor warehouse's on-hand quantity looks most comfortable. A fat on-hand number makes people careless, and one large order empties the floor. A healthy transfer is not "the receiving warehouse gets whatever it asks for." It is "the donor warehouse keeps enough, then gives what is left." The donor warehouse is not an infinite supply closet. It has its own orders to fill and its own customers to answer to. A transfer is a loan between two warehouses, not a one-way requisition.

So the next time someone brings you a transfer order to sign, ask one question first: how much is the donor warehouse keeping? If nobody can answer that, do not sign yet.

By Lin Xiao, WMS Implementation Consultant