Seller Fulfilled Prime, and When a 3PL Makes Sense

Most brands start on FBA and stay there, and for most of them that is correct. Amazon holds the stock, picks it, ships it, handles the returns and puts the Prime badge on the listing, and the badge is the part that matters: it is a conversion difference large enough that competing without it is a different business.

What changes with size is the arithmetic underneath it. FBA charges per unit for fulfilment and per cubic foot per month for storage, so a light expensive item is cheap to keep there and a heavy bulky one is not. At small volumes the difference is noise. At a few thousand units a month it is a line somebody notices.

It is the arrangement where a seller ships their own orders and keeps the Prime badge, and it is not a setting. It is a performance bar with a trial period attached, and the bar is what catches people.

The commitments are the ones a warehouse either meets or does not: shipping on time, at Prime speeds, with tracking that scans, at a cancellation rate close to zero, using approved carriers, including at weekends. A brand that ships four days a week does not qualify by trying harder. Losing the badge afterwards is worse than never having applied, because the listing has been converting at Prime rates and stops.

So the honest test is whether the fulfilment operation is already running at that standard for its own reasons. If it is, SFP formalises something true. If the plan is to raise the standard in order to qualify, the plan has the order wrong.

Three situations, and they are all about the shape of the product or the shape of the catalogue rather than about volume alone.

Bulky or heavy items, where FBA storage and the size band surcharges make the unit economics worse than a warehouse charging by pallet. This is the clearest case and the arithmetic is usually not close.

Slow moving range, where long term storage charges accumulate on stock that was ordered optimistically. A 3PL charging for space does not escalate the way marketplace storage does the longer a unit sits, and the escalation is what turns a bad forecast into a compounding cost.

Selling in more than one place, which is the one people underestimate. Stock committed to FBA is stock that cannot fill an eBay order or a Shopify order without being removed first. One pool of inventory serving every channel is worth real money in avoided stockouts, and it is invisible in any single channel's numbers. Running a catalogue across several marketplaces has its own problems, which we cover in repricing across marketplaces.

Small light products with steady velocity are the case FBA was designed for, and a 3PL rarely beats it. The per unit fulfilment fee is competitive, the storage is trivial because the cube is small, and the returns handling is included rather than being a line item somebody has to negotiate.

New products are the second case. A launch needs the badge, the ranking benefit and the delivery promise from day one, and it does not need an operational experiment running underneath it. Launch on FBA and reconsider once the demand is real, which is also the sequence campaign structure for a launch assumes.

And any seller whose current fulfilment is a person in a unit doing their best. That is not a criticism, it is a scale observation: Prime speeds at weekends with near zero cancellations is an operation rather than an effort.

Not one answer, which is the thing nobody tells you at the start. The common arrangement is fast moving small items in FBA, bulky and slow lines in a 3PL, and the 3PL serving every other channel from the same pool.

That is more complex to run and it is the arrangement the numbers usually point at, because the products genuinely differ. The mistake is deciding fulfilment once, as a policy, rather than per product line as a cost question.

The reporting has to follow it. A brand running two arrangements needs contribution margin per product after the actual fulfilment cost of that product, not a blended figure, because the blend hides exactly the lines that are losing money. That is the same discipline as reading what it costs to sell on Amazon rather than reading revenue.

Fulfilment cost per unit per product, including storage over the time that unit actually sits, rather than the headline fee. Return rate per product, because a 3PL quote that excludes returns handling is not a comparable quote. And the delivery promise each option produces in each region, because that is a conversion input rather than a service detail.

Then the switching cost, which is real and is usually left out: removal orders, freight to the new warehouse, the integration work, and a period where both are running.

None of this is a reason not to move. It is a reason to move a product line at a time, measure it, and let the second one be an easier decision than the first. Getting that sequencing right across a catalogue is a large part of what Amazon account management is for.