PPPick Path Notes
Issue 21  dock to promise · eight sequences2026-11-20
PREP  /  Note 06

Prep is a specification with a rejection clause, and that changes how you staff it

Amazon prep and third party prep look like the same work performed in different buildings. The difference is that one of them has a published specification with an enforcement mechanism, and the other has a conversation.

LanePrep
Sections10
Checks12
Filed2026-11-20
Amazon prep and third party prep look like the same work performed in different building
Amazon prep and third party prep look like the same work performed in different buildings. The difference is that one of them has a published specification with an enforcement mechanism, and the other has a conversation.

Who owns the prep decision

In an FBA flow, the prep specification is written by the marketplace and the merchant is the party that eats the consequence of getting it wrong. In a third party flow, the specification is written by the merchant and the warehouse executes it. The ownership difference explains most of the behavioural difference.

Marketplace specifications are enforced by rejection. A non compliant unit is refused at the inbound dock, and the refusal carries a fee, a delay and a return leg. There is no negotiation at the point of refusal, which is why the specification has to be read before the shipment is built rather than after it is refused.

Merchant written specifications are enforced by the service agreement, which means they are enforced only to the extent that the agreement is specific and the reporting is honest. A warehouse that is told to poly bag everything will poly bag everything, and it will not tell the merchant that half the catalogue does not need it.

The staffing consequence is direct. FBA prep needs a compliance role that reads specifications and audits output. Third party prep needs a specification role that writes them and a reporting role that proves they were followed. Those are different jobs and they are rarely both staffed.

The FNSKU label covers rather than accompanies

An FNSKU label has to cover the manufacturer's barcode, not sit beside it. Two scannable codes on one unit is a unit that will eventually be scanned with the wrong one, and the wrong scan puts the unit in the wrong inventory pool.

Covering is a physical act with a quality dimension. A label that covers the barcode but leaves a corner of the original visible will still read as the original under some scanners. A label applied over a curved surface without a squeegee will lift at the edges and jam the sorter.

Label placement is specified by the marketplace, and the specification is about the scannable area rather than the unit. The label needs clear space around it, no curvature across the bars, and no tape over the bars. Tape over a barcode produces a specular reflection that defeats the scanner.

The audit for this is a scan test rather than a visual test. Scanning each unit with the actual warehouse scanner, at the angle the sorter uses, catches the placement errors that a visual inspection passes. A visual pass rate of ninety nine percent can coexist with a scan pass rate of ninety percent.

Poly bags are specified by opening size, not by product size

The suffocation warning requirement is triggered by the bag's opening dimension rather than by the product it contains. A small product in a large bag carries the warning requirement because the bag is the hazard. A large product in a tight bag may not.

The threshold is five inches of opening, and the warning has to be printed on the bag or on a label affixed to it, in the destination language, in a legible size. A warning printed on the product's own label rather than on the bag does not satisfy the requirement.

Bag thickness is a separate requirement with a separate failure mode. A bag below the minimum thickness tears at the seal and produces a unit that arrives loose, which is a condition complaint rather than a prep rejection. The two specifications are often confused because they are both about bags.

Air removal matters for dimensional weight as well as for compliance. A bag with trapped air is a bag that inflates the parcel and can push it into a higher dimensional band. Vacuuming or pressing before sealing is a freight saving that happens to also be a compliance step.

Sets and bundles have to be unsellable as parts

A bundle is defined by the fact that its components cannot be separated and sold individually. That is a physical property rather than a labelling property, and it is the property that most bundles fail.

The physical requirement is that the components cannot be removed without destroying the packaging. A bag containing three items that can be pulled out one at a time is three items in a bag, not a bundle, and the marketplace will treat it as three items if it is discovered.

The labelling requirement is that the bundle carries a single scannable identity that resolves to the bundle rather than to a component. A bundle whose label resolves to the largest component will be received as that component and will corrupt the inventory record.

Bundling also changes the dimensional weight and the prep labour per unit. A bundle that is three units is three times the prep time and often more than three times the volume. The pricing has to reflect both, and it rarely does when bundles are created at the listing stage without warehouse input.

Expiry and lot data is a field, not a sticker

Categories with shelf life require the expiry date and the lot to be readable on the unit. A sticker with a handwritten date satisfies the visual requirement and fails the data requirement, because the marketplace expects the data in the shipment record as well as on the unit.

The lot is what makes a recall possible. A recall that can only identify the shipment rather than the lot cannot distinguish the affected units from the unaffected ones in the same shipment, and the resolution is to destroy the whole shipment.

The print quality requirement is real. A date printed by a thermal printer that is fading will be unreadable by the time the unit reaches the customer, and unreadable expiry data is treated as missing data. The printer's heat setting and the label stock are a specification, not a consumable choice.

Rotation is the consequence. Units with expiry have to be picked first expired first out, which requires the date to be in the warehouse system rather than only on the unit. A warehouse that stores the date only on the sticker cannot rotate, and the write off is the evidence.

The per unit fee structure decides which prep is cheaper

Prep fees are usually quoted per unit and broken down by operation: label, poly bag, bubble wrap, bundle assembly, boxing. Comparing two providers on a single blended per unit rate hides the mix, and the mix is where the difference is.

The relevant comparison is the total prep cost for a representative basket of the catalogue rather than an average rate. A provider that is cheap on labelling and expensive on bundling will look attractive against a catalogue that is mostly single units and expensive against one that is mostly bundles.

There is also a minimum charge structure that dominates at low volume. A monthly minimum means the effective per unit rate at two hundred units is several times the quoted rate, and the quoted rate is what gets compared.

The hidden line is the rejection cost. A prep provider whose output is rejected at inbound has a per unit cost that includes the rejection fee, the return leg and the reship. That cost does not appear on the quote and it appears in the monthly total.

The three reasons shipments get refused

In practice most inbound refusals come from three causes: an unscannable label, a bag below the thickness minimum, and a carton over the weight limit. All three are detectable before the shipment leaves and all three are routinely missed because the check happens on a sample rather than on every unit.

Unscannable labels are usually a placement problem rather than a print problem. The label printed correctly and was applied across a curve, or under tape, or beside the original barcode. The fix is a placement jig and a scan test at the end of the line, not a new printer.

Bag thickness is measured with a micrometer at the seal, not at the centre. A bag that measures above the minimum at the centre and below it at the seal will pass a casual check and fail a proper one. The seal is where the thickness collapses.

Carton weight limits apply to the shipment carton rather than to the unit, and they are lower than most people assume. A carton packed to what feels like a sensible weight for a person to lift can exceed the marketplace limit, and the refusal happens at the dock after the freight has been paid.

Third party prep is more flexible and less auditable

A third party warehouse can change a prep specification in an email, which is an advantage when a product changes and a disadvantage when the change is not documented. The flexibility is real and it comes with an audit gap.

The audit gap is closed by asking for evidence rather than for assurance. A monthly sample of photographs per prep operation, a scan test result, and a count of units prepped against units received. Those three artefacts make the claim testable.

FBA prep is less flexible in the sense that the specification cannot be negotiated, and it is more auditable in the sense that the enforcement is external. The trade is real and it should be made per category rather than as a company wide preference.

A hybrid that works: third party prep for the categories where the merchant needs a bespoke insert or a bundle that the marketplace specification does not describe, and FBA prep for the commodity units where compliance is the only requirement.

One SKU can serve two channels at once

A SKU does not have to choose between FBA and third party fulfilment. It can ship a share of its receipts to the marketplace and keep the rest in the third party warehouse, and the split can be adjusted without changing the supplier order.

The split changes the prep requirement, because the marketplace share needs FNSKU labels and the third party share does not. Labelling everything is wasteful and labelling only the designated share requires the split to be decided before the cartons are opened.

The economic driver of the split is the storage cost curve against the fulfilment fee curve. Marketplace storage becomes punitive after a certain age, and third party storage is a flat rate per pallet. The crossover point is computable per SKU and it moves with the sales velocity.

The risk driver is single channel dependency. A SKU that exists only in one channel is a SKU that disappears when that channel has an issue. The split is therefore partly an insurance decision, and the insurance is worth more for the SKUs that carry the revenue.

Comparing the two on a like for like basis

A like for like comparison has to include four lines that quotes usually omit: inbound freight to each location, storage cost at the actual dwell time, prep cost at the actual mix, and the cost of exceptions including rejection.

Inbound freight is the line that most often reverses the conclusion. A third party warehouse in a different region has a different inbound cost, and a marketplace inbound has its own placement requirements that may force split shipments to multiple nodes.

Dwell time is the assumption that decides the answer. A merchant that models storage at thirty days will find one answer and a merchant that models it at ninety will find another, and the right assumption is the one from the sales history rather than the one from the plan.

Exceptions should be modelled as a rate rather than as a risk. The rejection rate at a marketplace inbound is measurable from the account, and the prep error rate at a third party warehouse is measurable from the audit sample. Both numbers exist and both belong in the model.

The provisioning side of this comparison, covering the labels, the bags and the per unit fees, is the marketplace prep alongside third party prep, labelling and poly bagging, and the fee breakdown is what makes a like for like comparison possible.

Checklist12 lines
  • 01Read the marketplace prep specification before the shipment is built, not after it is refused.
  • 02Apply FNSKU labels over the manufacturer barcode, never beside it.
  • 03Scan test every unit at the sorter angle instead of inspecting visually.
  • 04Check the poly bag opening dimension against the warning threshold.
  • 05Measure bag thickness at the seal rather than at the centre.
  • 06Make bundles physically inseparable rather than merely bagged together.
  • 07Record expiry date and lot in the system, not only on the unit.
  • 08Compare prep providers on a representative basket rather than a blended rate.
  • 09Add the rejection fee and the return leg to the per unit prep cost.
  • 10Verify carton weight against the marketplace limit before tendering freight.
  • 11Require photographic and scan test evidence from a third party prep provider.
  • 12Decide the channel split per SKU from the storage and fee crossover point.