Warehouse SOP templates: receiving, picking and counts

The short answer

Five procedures cover most of what goes wrong in a small warehouse: receiving, putaway, picking, packing and cycle counts. Written properly, each one names who does it, what gets checked, and what happens when the check fails. That last part is the one people skip, and it is the only part that matters at eight in the morning during peak.

Larger operations run fifteen to forty procedures covering dispatch, returns, stock adjustments and safety. A brand shipping a few hundred orders a week does not need forty. It needs five that are actually followed.

Which of three situations you are in

The middle row is the one nobody writes about, and it is where most DTC brands actually sit. If a 3PL runs the floor, you do not write their picking procedure. They have one. What you write is the procedure for the seam between you and them, and that section is further down.

Receiving and putaway

Receiving is where inventory accuracy is won or lost. Every count error later in the year traces back to a pallet that got waved through because the delivery arrived at a bad moment.

1. Check the delivery against the purchase order before the driver leaves. Cartons, not units. If the carton count is wrong, note it on the delivery paperwork while the driver is still there, because a discrepancy raised afterwards is a conversation you lose.
2. Photograph any damaged or opened carton before opening it. This is your evidence for a supplier claim and it takes ten seconds.
3. Open and count units against the packing list for at least one carton per SKU, and every carton for a first order from a new supplier.
4. Check the product itself, not just the count: right colourway, right size run, right expiry or batch where it applies.
5. Record the received quantity in your system the same day. Not at the end of the week. Stock that exists physically and not in the system will be oversold.
6. Put away to a fixed location and record it. A SKU that lives in a different place each delivery is a picking error waiting to happen.
7. File the paperwork against the purchase order so finance can match invoice to receipt later.

Picking and packing

Picking errors are expensive twice: once for the wrong item shipped, again for the return, the reship and the support ticket. A wrong-item rate of even one percent is a support load you will feel.

The picking procedure:

1. Pick from a printed or on-screen pick list, in location order rather than order-of-arrival. Walking the same route every time is what makes the job fast.
2. Scan or check the SKU against the pick list at the shelf, not at the packing bench. Catching it at the shelf costs seconds; catching it at the bench costs a re-walk.
3. For multi-item orders, pick into a tote with the order number visible. Loose picking is how two orders get merged.
4. Flag a short pick immediately rather than substituting. A substitution nobody recorded turns into a customer complaint and an inventory discrepancy at once.

The packing procedure:

1. Verify contents against the packing slip before sealing. This is the last checkpoint before it becomes a customer problem.
2. Photograph high-value orders before sealing. It settles a not-as-described dispute in one reply.
3. Include whatever the marketing team has agreed goes in the box, and keep that list somewhere they can update without asking you.
4. Weigh and record if your rates are weight-banded, since a wrong weight is a billing correction later.
5. Mark the order shipped in the system at the point of dispatch, not at the point of packing. The gap between the two is where tracking complaints come from.

Cycle counts

A full annual stocktake tells you your inventory was wrong for a year. Cycle counting means counting a slice regularly, so an error surfaces within weeks rather than at year end.

1. Segment by value and velocity. Your top-selling and highest-value SKUs get counted monthly. The long tail gets counted quarterly or twice a year.
2. Freeze movement for the SKUs being counted, or count outside picking hours. Counting a SKU while someone picks it produces a discrepancy that is not real.
3. Count blind. Do not show the counter the expected number, because a counter who can see the target will find the target.
4. Recount any variance before adjusting. Most first-count variances are counting errors, not stock errors.
5. Adjust in the system with a reason code, not a bare number. Damage, theft, miscount and receiving error are four different problems with four different fixes.
6. Track variance by SKU over time. A SKU that is repeatedly short is telling you something specific, and it is rarely theft.

Counting blind is the step teams argue about, because showing the expected figure is faster. It is faster and it is worthless. The purpose of the count is to discover what is there, and a number on the sheet is a strong suggestion.

When a 3PL runs the floor

If a third party fulfils your orders, most of the procedures above belong to them. Writing your own version of their picking process achieves nothing, because nobody in that building will ever read it.

What you write instead covers the seam. Four procedures, and they are the ones that break:

• Sending inventory in. What paperwork accompanies a shipment, how SKUs are labelled, how you tell them a new SKU exists before it arrives. An unannounced SKU sits in receiving for a week.
• Checking what they received. Their receipt against your purchase order, and who chases the difference. This is the single most valuable procedure in the list, and it is the one most brands do not have.
• Handling exceptions they escalate. Damaged units, missing cartons, a customer address they cannot ship to. Who decides, and within how long.
• Reconciling stock. How often you compare their numbers against your storefront, and what you do when the two disagree.

Ask for their SOPs during onboarding and read them. Not to rewrite them, but so you know what they actually check on receipt, because that determines what you have to check yourself.

What actually breaks

Three failures account for most of it, and none of them is solved by writing a longer document.

The first is seasonal staff. You write procedures in March for a team that knows the job, then hire four temps in October who have never seen the building. A procedure that assumes context is worthless to the person it was written for. Test it by handing it to someone who has never done the task and watching where they stop.

The second is that the procedure describes a system that has since changed. Your inventory tool ships an update, the screen moves, and the written step is quietly wrong. Nobody announces it. You find out when a new hire gets stuck.

The third is that no procedure has an owner. An unowned document is nobody's job to fix, so it degrades at a predictable rate until people stop trusting the folder entirely, at which point you have the cost of documentation and none of the benefit.

Start with receiving. It is the procedure with the most downstream consequences, it takes an afternoon to write, and its failures are the ones that show up months later as an inventory number nobody can explain.