The FBA Shipment Checklist: Unit, Box, Pallet, Paperwork

Every FBA shipment has to satisfy the same base layer, whatever is inside the boxes: unit identity, packaging that survives handling, carton build, pallet build, and the data that travels with the freight. Category law stays out of it on purpose. Hazmat paperwork, apparel bagging, meltable season windows, supplement dating: each carries its own thresholds and its own guide. Clear the general rules first, because a category rule satisfied on a unit that fails the general ones is wasted work.

Run it with the boxes still open and the tape gun still on the bench. Every line answers yes or no. The ones answered probably fine are the ones that come back as an inbound defect three weeks later, when the units sit in a fulfillment center and nobody on your side can reach them. Adding a missing bag on your own floor takes two seconds. Adding it after a truck has the pallet takes a removal order.

Two dates that moved the work to you

On January 1, 2026, prep and item labeling stopped being services Amazon sells for US FBA shipments, including inventory routed through AWD, AGL, SEND and the Supply Chain Portal. Sellers on the Fulfillment Inbound API saw it in the schema before the announcement: AMAZON is no longer an accepted value for prepOwner or labelOwner in the US.

The effect lands at receiving. A unit without the bag it needed, or without a barcode anyone can scan, no longer gets corrected on Amazon's side. It waits, goes back, or draws an inbound defect charge, and sells nothing the whole time.

The second date is March 31, 2026, when stickerless commingled inventory ends. Settle it before anything prints: read what the end of commingling does to resellers, work out which side of that line your account sits on, and write down who has to put an Amazon barcode on every piece. A barcode decision made after the labels print costs the whole run.

The unit a customer will open

Start with one sellable unit in your hand. Not the case pack, not the master carton. Finish this level before moving up: packers who bounce between the bench and the pallet skip a line and hear about it at check-in.

Labels get one line here, and it is a hard one. Exactly one barcode is scannable from outside the unit, and it resolves to the SKU on the shipping plan. Print quality, thermal fade, the second barcode nobody covered: those are the label failures receiving actually catches, and each one turns a good unit into an exception.

Bagging and dates outgrow a table cell. Bagging carries thresholds of its own, from which openings need a suffocation warning to whether a scanner reads the code through the film (poly bag rules, opening by opening). Dates have to be numeric and printed, and a lot code is not one, so check the format and the remaining shelf life against Amazon's expiration date rules before a pallet gets built on top of them.

If your units arrive already prepped by a supplier, this level moves earlier rather than disappearing: somebody opens cartons and checks a sample before the load is committed to a shipping plan. Catching a bad label run in a sample costs one carton of time. Catching it after the pallets are inside Amazon's building costs the shipment.

The carton

Now the shipping box. It plays to two audiences: the conveyor that will throw it around, and the scanner at receiving.

1. Box condition. Rigid, six-sided, corrugated, meeting the burst strength or edge crush rating Amazon names in its packaging requirements. Fails on crushed corners, cutouts, handle holes, or a grocery case with the top flaps torn off.
2. One shipment per box. Everything inside belongs to the same shipment ID. Fails the moment two shipping plans share a carton, even by one unit.
3. Cushioning. Bubble wrap, air pillows, or heavy kraft paper. Fails on packing peanuts, shredded paper, crinkle paper, and foam strips. Those are refused materials, not preferences.
4. Shake test. Lift the sealed box and shake it. Nothing shifts, nothing rattles, no unit sits directly against a wall. Fails if you can hear the contents move.
5. Size and weight. Inside the box dimension and weight ceilings Amazon sets for inbound cartons, unless the box holds a single oversize item that cannot be split. Fails when a heavy SKU quietly pushed a carton over the line and nobody reweighed it.
6. Handling labels. Heavy, team lift, and mechanical lift markings applied when the weight band calls for them. Fails when a carton at the top of the range carries nothing warning the person lifting it.
7. Old labels off, new labels on. Every previous carrier label, pallet label and supplier barcode removed or blacked out; the FBA box ID label and the carrier label flat on a face, neither bridging a seam nor overlapping the other. Fails if an old tracking barcode is still scannable anywhere on the box.
8. Tape. Pressure-sensitive plastic tape or water-activated reinforced tape across all seams. Fails on masking tape, duct tape, string, and staples anywhere near a seam.

Box content information deserves its own paragraph, because sellers file it under paperwork when it belongs to the box. Manual entry, file upload, a 2D barcode for the receiving team to scan: the method does not matter. The declared SKUs, quantities, weights and dimensions have to match what is physically inside that specific carton, not what is true of the shipment as a whole. Split a carton at the last minute and the declaration gets redone before the labels go on. A shipment arriving with no box content information picks up a per-unit manual processing charge, and a pattern of mismatched cartons can put later shipments under restriction. The 2D barcode is the version that scales: receiving scans one code per carton instead of counting units.

The pallet

Small parcel shipments skip this level. On LTL, full truckload, or a container, the pallet is a unit of compliance in its own right: it is accepted or refused as one object, not one carton at a time.

1. The pallet itself. Wood, four-way entry, the standard grocery footprint and grade named in the FBA requirements, no broken deck boards, no protruding nails. Fails on plastic pallets and on anything a forklift can only enter from two sides.
2. One destination per pallet. Every box on it belongs to the same shipment going to the same fulfillment center. Fails when two destinations share a stack because it was easier to wrap once.
3. Stacking. Flush on all four sides, no overhang, heavier cartons at the bottom, columns square instead of creeping toward the deck edge. Fails if a carton hangs over the pallet by any amount.
4. Label orientation. Every box ID label faces outward and can be scanned without cutting wrap or breaking the stack down. Fails when an inner column of labels points into the middle of the pallet.
5. Height and weight. Under the pallet ceilings Amazon sets for inbound freight, measured with the pallet included. Measure the stack and forget the deck, and the load gets rebuilt on somebody else's dock.
6. Wrap. Clear stretch wrap, carried down over the pallet base so the load and the deck move together. Fails on black or colored wrap, and fails if the top layer can be pushed sideways by hand.
7. Pallet labels. One on each of the four sides, applied on top of the wrap, flat and fully visible. Fails when a label goes under the wrap, gets taped to a carton, or ends up folded around a corner.
8. Stacked or not. Build single-stacked unless the freight was booked as stackable and the cartons underneath can carry a second pallet. A pallet crushed by the one riding above it becomes a damage claim nobody wins.
9. Photo. Two angles of each finished pallet before it leaves the dock, plus a shot of the scale reading. Not an Amazon requirement, and exactly what you will want if a carton goes missing later.

What gets a load turned away at the gate usually looks fine from ten feet away. Wrap that holds the cartons to each other but not to the pallet is one case: the stack arrives leaning, corners crushed, and the receiving team treats it as damaged freight. Carry the film down over the deck and give the base a few extra turns. A pallet label applied before the wrap goes on is the other: readable on your dock, invisible under three layers of film, and a pallet nobody can scan is a pallet nobody unloads.

Paperwork that has to agree with the freight

Paper is the level where nothing looks wrong until the driver is already at the gate.

• The shipment is still valid in Send to Amazon and the shipment ID on the labels matches the plan you are actually shipping against.
• Box content information has been submitted, and the carton count in the plan matches the carton count on the floor.
• The carrier is assigned in the plan. On a partnered carrier, the labels you printed came after the rate quote was accepted.
• The bill of lading names the correct destination and carries the shipment reference. On collect freight, the Amazon reference number is on it, valid, and unused.
• The carrier has booked a delivery appointment through Carrier Central against the right reference number and the real pallet count, and you have the appointment ID. Freight without an appointment does not get unloaded, and the booking locks in the last stretch before the window.
• Weights and carton counts on the bill of lading match your own record instead of an estimate somebody rounded.

When a line fails

A failed line turns into a crisis only when somebody invents a fourth option. There are three.

1. Fix it now. Relabel, rebag, repack, reweigh, then rerun that level from the top for the unit or carton involved. Half a fix is not a fix.
2. Pull it from the shipment. Take the units out, correct the quantities in the plan and in the box content data, ship the rest. A short shipment reconciles cleanly; one with phantom units does not.
3. Hold the whole load. When the failure is systemic, one bad label file or one wrong barcode across a production run, shipping it is the expensive choice. Rebooking a pickup costs less than a thousand units flagged at receiving.

If option one keeps landing on the same SKUs, the problem sits upstream of the checklist and no amount of care at the dock will fix it. That is usually the point where a seller hands the bench work to a prep center and stops rerunning label jobs at midnight; what that runs per unit is on the prices page.

Settled weeks before pickup day

Some lines cannot be checked at the dock, because they were decided long before it. Confirm them while changes are still cheap.

• Who preps and who labels. The answer is you or somebody you hire. Amazon is off the list.
• Which barcode each SKU carries. Manufacturer barcode or Amazon barcode, decided per SKU by your Brand Registry status and written down where the packing team can see it, not held in one person's head.
• Which placement option you are taking. Splitting the shipment the way Amazon asks sends you to more inbound locations, carries no placement fee, and opens only once a SKU meets the minimum quantity set for the split. Consolidating to one or two destinations is easier on your dock and costs the fee. Your pallet count follows that answer, so settle it before you build.
• How box content gets submitted. Manual entry works for a handful of cartons. Past that, set up file upload or 2D barcodes before the day of the shipment rather than during it.
• Who owns rework. If a supplier preps for you, agree in writing who pays when a load fails inspection, and agree before the load is on a boat. Cargo coming in from overseas puts a port, a broker and a domestic leg between you and the defect, which is why prep for international sellers gets planned around a checkpoint on this side rather than around the factory.
• Date policy on dated goods. The minimum remaining shelf life you accept from a supplier, first in first out on your side, and a standing rule for short-dated stock that arrives anyway.
• What outsourcing actually costs. Compare a quote against your own labor and your own error rate rather than against zero.

The log that becomes a purchasing decision

Keep the shipment evidence together while it is fresh: pallet photos, the scale reading, the signed bill of lading, the appointment ID, and the box content file as submitted. That folder decides how a later case goes, including whether a reimbursement lands anywhere near what the unit cost you. Shipments that get rejected, delayed, or surcharged follow patterns worth knowing before you have to open one.

The other habit worth building is a defect log: one line per problem caught during prep, with the date, the SKU, the supplier, and the check that caught it. Early on it reads like a list of annoyances. Ten shipments in, it names the supplier whose cartons always need rework and the SKU whose bags always split, which makes it a purchasing decision with evidence attached.

Asked at the packing bench

A shipment already left with unlabeled units. What now?

Speed decides the cost. If the carrier has not delivered yet, recall the shipment and finish the units on your own floor. Once they are received, nobody stickers them for you: they sit unsellable until you pull them back with a removal order, relabel off site, and build a fresh plan for the same units. Price that removal freight before you call this a small problem.

How do I check which barcode a specific ASIN needs?

Check it per ASIN, because one seller can hold different roles across a catalog. Open the listing from Manage Inventory and read the barcode setting on the offer, then read the labeling owner on that SKU line in your shipping plan before anything prints. If the two disagree, follow the shipping plan for this shipment and fix the listing setting afterward.

How much gets checked on every unit and how much on a sample?

Anything that scans gets checked on every unit, because one wrong label is enough to create a mismatch nobody finds until receiving does. Condition, bagging and dates can run on a sample when the units come from a single production run and the sample is pulled from several cartons instead of the top one. Cartons and pallets get checked in full; there are few enough to make sampling pointless. The moment a sample turns up one defect, stop sampling that batch.

What happens if box content information does not match the carton?

Receiving counts what is physically there. A mismatch slows check-in and can trigger a manual processing charge; a pattern of them puts later shipments at risk of restriction. Correcting the plan on your own dock beats arguing about a count afterward.

My carrier labels the freight. Do I still need FBA pallet labels?

Yes, and they do different jobs. Carrier paperwork moves the pallet between docks; the FBA pallet label tells the receiving building which shipment the stack belongs to, which is why it goes on all four sides, on top of the wrap. A stack labeled on the front only gives the unloading team a reason to set it aside.

Can I reuse boxes from my supplier?

Yes, if the box is structurally sound and every old label and barcode is removed or completely covered. Crushed corners, cutouts, missing flaps and stray tracking barcodes are why reused cartons fail. A scannable old label is worse than a scuffed box: it sends the carton somewhere instead of just looking bad.

Does any of this change if a prep center packs the shipment?

The levels stay the same; the ownership moves. A contractor runs the unit, carton and pallet checks and signs off on them, which is what you are paying for. The decisions above the checklist stay yours: the barcode each SKU carries, the placement option, the shelf life you accept from a supplier. Hand those over as written instructions, ask for the pallet photos back, and check the per-unit rates on the price list before comparing it to your own bench.