Shipping From Someone Else's Warehouse? The Paperwork That Decides Who Eats a Shortage

The decision to keep stock in someone else's building is usually easy to justify and easy to underestimate. Rent gets replaced by a per-pallet storage fee, staff by a pick charge, and a lease by a contract you can leave in ninety days. What does not disappear is the count. Every unit you own now sits in a facility you cannot walk into on a Tuesday afternoon, which means the record of what arrived, what shipped and what went missing has to be built out of documents rather than eyesight. That record is the whole job.

The inbound notice is the document that decides everything after it

Goods arriving at a third-party warehouse without an advance shipping notice get received against whatever the driver hands over, which is often a carrier's bill of lading listing cartons rather than contents. That is where discrepancies begin, because a receiving clerk counting eighteen cartons has confirmed eighteen cartons and nothing about the four hundred units inside them. Send the notice before the truck: purchase order number, carton count, units per carton, SKU per carton, expected arrival window, and the carrier's pro number. A facility that can match a delivery to an expectation will tell you within a day that it is short. One that cannot will tell you in six weeks, when you are out of stock.

Carton labeling belongs to the same handover. If the warehouse has to open boxes to work out what is in them, receiving slows down and errors get introduced by people who have never seen your product before. Label the outside with the SKU, the quantity, the purchase order and a barcode that resolves to your own item number rather than a manufacturer's. Ask your supplier to do it at origin and put it in writing on the purchase order, because a factory that labels correctly the first time saves you a repackaging charge that shows up on a bill you did not expect.

SKU discipline is cheap now and expensive later

Operators who have moved inventory between two or three providers all say the same thing about item numbers: fix them before the first pallet ships, because renaming a SKU inside a live warehouse management system usually means a new item record, a physical relabel, and a period where two numbers describe the same thing. Use one SKU per sellable unit, with a separate number for each case pack and each bundle, and never reuse a retired code for a new product. Keep a master list you own, in a spreadsheet or a system you control, not only inside your provider's portal.

Barcodes matter more than the naming convention. A scannable code on every sellable unit and every case turns receiving, picking and cycle counting into machine work with an error rate you can actually measure. Products that arrive without one get a warehouse-applied label at a per-unit fee, which is fine occasionally and ruinous across a container. Where an item is sold in a multipack, make sure the multipack carries its own distinct code, because a picker scanning the inner unit six times is not the same transaction as a picker scanning one bundle, and the inventory ledger knows the difference.

Read the rate card as a list of behaviors, not prices

Storage and pick fees are the numbers people compare, and they are rarely the numbers that move a monthly invoice. The accessorials do that: receiving by the hour versus by the pallet, minimum storage charges per SKU, pallet in and out fees, special project labor, repackaging, returns processing, long-term storage surcharges after six or twelve months, and a per-order administrative charge that quietly attaches to everything. Ask for a sample invoice built from your own forecast volumes, with each line named. Any provider who has done this before will produce one without hesitation, and the exercise tends to reveal which of your own habits are costing money.

The rate card also encodes what the operation is good at. Hourly receiving usually means mixed, unlabeled freight is normal there. Aggressive per-SKU minimums mean the facility is built for fast movers and will penalize a long tail of slow ones. Cheap picks with expensive projects means the automation is real but anything unusual gets handled manually. Match those signals to your catalog rather than to the headline rate, and you will pick a provider whose economics point the same direction as yours.

The systems handshake, and the reconciliation nobody sets up

Orders reach the warehouse through an integration, and inventory comes back through the same pipe. Whether that is an EDI connection, a direct API, or a middleware connector sitting between your store and their management system, the thing to establish on day one is what happens when a message fails. Confirm who is notified, how quickly, and whether a rejected order sits in a queue somebody reviews or in a log nobody opens. Ask to see the error report format. The integration will work most of the time; the value of the arrangement is decided by what happens on the days it does not.

Then set up a cycle count schedule and read the results. A weekly or monthly count of a rotating slice of SKUs, reconciled against your own record of receipts, shipments and returns, catches drift while it is still small enough to explain. Agree in the contract on a shrink allowance, a claims window, and the evidence required: a signed proof of delivery, dated photographs, the receiving report. Providers with mature operations expect this conversation, and the ones who propose the counting cadence themselves are generally the ones worth signing.

Outbound records protect the sale, not just the stock

Every shipment should generate a tracking number written back to the order, a packing slip that matches what left the building, and a timestamp you can retrieve months later. That trio settles chargebacks, answers customers, and supports insurance claims for freight lost in transit. It also keeps you on the right side of shipping-time representations to buyers, an area the Federal Trade Commission oversees for mail, internet and telephone orders, where the promise made at checkout has to line up with what the warehouse can actually do on a given day.

Returns need the same treatment in reverse. Decide in advance what gets restocked, what gets scrapped, and who authorizes the difference, then require a disposition record per unit rather than a monthly total. Inventory that comes back into a saleable bin without inspection becomes a second complaint from a second customer. Handled properly, a returns line item that used to read as pure loss turns into recovered stock with a documented reason code, which is also the input you need to fix whatever caused the return.

None of this requires a large system. A purchase order template with labeling instructions, a SKU master you control, an annotated rate card, a cycle count calendar and an agreed claims process cover almost every dispute that arises between a seller and a warehouse. Set them up while the relationship is new and cooperative, because that is when both parties are willing to write down what they will do.