A guest post, and what that means here. This one is not ours and not about tax. It was written by the FLEX. network, a European third-party logistics provider with warehouses in Germany, France, Poland and the UK, and it covers the operational half of a German launch: where the stock lands, who receives it, and what has to be agreed before the freight moves. We publish it because our customers keep asking these questions after their VAT number arrives, and because it is a better answer than ours would be. Vaytax does not provide fulfilment, warehousing or logistics services, and nothing here is tax advice. The companion piece, on the VAT side, was written by our licensed German tax advisor for FLEX.'s own site.
The number comes through, the registration file closes, and Seller Central stops warning you about the country you were trying to enter. The obvious next action is to book the freight.
There is a gap between those two events, and a large part of what makes a German operation cheap or expensive, flexible or rigid, gets decided inside it, by choices that are straightforward to make in advance and awkward to unwind once inventory has landed.
“Germany” is not an inbound plan
A destination country is not a destination. Before the freight is booked, the inventory needs a specific first address, and there are three common shapes.
Direct: supplier or import point straight into an Amazon fulfilment centre. With a layer: supplier or import point into an independent German warehouse, then onward into Amazon. Or wider: a German warehouse that serves both Amazon replenishment and direct B2C and B2B orders.
These are not three grades of the same thing. The first commits the inventory to one fulfilment network the moment it is received. The second keeps a decision point between arrival and commitment. The third makes the stock available to channels that are not on the plan yet.
The right shape depends on the channels in play, replenishment frequency, SKU count, inventory depth, prep requirements, expected returns, whether reserve stock is needed at all, and whether the seller wants inventory reachable outside Amazon.
Routing sits underneath all three: where the goods clear, who is named on the booking, who arranges the leg from the clearance point to the warehouse, and what delivery window that warehouse accepts. Sellers who leave this to the forwarder alone tend to discover at the dock that the receiving site expected something else.
Direct to Amazon, or a German 3PL first
This is a decision framework rather than a ranking. Neither model wins on principle.
Sending direct to Amazon works well when:
- the inventory is already fully compliant on arrival;
- Amazon is the primary fulfilment channel;
- inbound quantities and replenishment are predictable;
- little has to happen between manufacture or import and Amazon receipt;
- there is no real requirement for inventory outside the FBA network.
Under those conditions a warehouse layer adds a handling step and a cost line without adding control.
A German warehouse layer earns its place when:
- stock needs inspection or preparation before it can enter FBA;
- inventory arrives in quantities larger than should sit inside Amazon at once;
- the seller wants buffer stock held outside the network;
- replenishment should happen in smaller batches;
- the same inventory has to serve more than one channel;
- Amazon removals need a destination and returned units need inspection or rework;
- there is relabelling, re-boxing or bundling to do.
The framing that misleads people is “3PL versus FBA”. In practice they are usually complementary layers: the 3PL acts as the inventory and control layer, and Amazon FBA is one of the fulfilment destinations that layer feeds.
Receiving: someone has to own the handover
Receiving is the first place the plan meets reality, and it is not simply a truck arriving. The warehouse needs to know what is coming before it comes: expected quantities by SKU, whether goods arrive palletised or as loose cartons, carton counts and units per carton, and how each SKU can be identified on arrival. That is what an advance shipping notice is for, and its quality decides how quickly received stock becomes visible and usable rather than sitting in a query queue.
Then there is the exception set, which is where the cost lives. Quantities that do not match the booking. Cartons damaged in transit. Units damaged inside intact cartons. Missing, wrong or unscannable labels. SKUs arriving mixed in a carton the paperwork described as single-SKU.
None of this is unusual. What varies is whether the process was agreed beforehand or improvised at the dock. Define receiving before dispatch: who is authorised to decide what happens to a discrepancy, and how quickly the seller hears about it.
Decide where preparation happens
Stock that is commercially ready to leave a factory is not automatically ready for every fulfilment channel. Something has to happen in between, and the useful question is where. That work can include inspection, product labelling, FNSKU labelling where the SKU requires it, bagging or protective packaging, bundling and multipacks, kitting, re-boxing, carton preparation, palletisation, and channel-specific requirements that differ between a marketplace shipment and a parcel to a consumer.
Marketplace packaging requirements change and should be checked against current documentation rather than an article. The strategic point does not: if preparation cannot happen reliably at origin, it has to happen somewhere in Germany before the goods become sellable, and choosing that location is a design decision. Prep done at the factory is cheapest when it is correct and most expensive when it is not, because the correction lands after the freight has been paid for.
Buffer stock changes the replenishment problem
There is a meaningful difference between all available inventory sitting inside Amazon from day one, and reserve inventory held outside the network with controlled replenishment into FBA.
The first is simpler. It also commits every imported unit to one fulfilment network at the moment of receipt, at whatever the forecast happened to be when the container was booked. If velocity is lower than expected, the stock is in the wrong place and comes back out through a removal.
The second keeps the units in a position from which they can still be allocated. Stock is released into Amazon in batches sized against actual velocity, the remainder stays available for other channels, and inventory is not concentrated in one network before there is evidence for it. It costs an extra handling step per batch, and buys the ability to redirect inventory before it is committed.
Sellers who want to hold reserve inventory outside Amazon need a layer capable of receiving, storing, controlling and releasing stock as the replenishment plan changes. FLEX. runs warehouse operations in Germany as part of a wider European 3PL network, alongside sites in Poland, France and the UK.
Replenishment is a process, not another inbound shipment
Once inventory is split between warehouse stock and Amazon stock, replenishment stops being an event and becomes a loop that needs owners and rules. The operating model should answer, in advance: what signal triggers a replenishment, which SKUs move and in what quantity, where any remaining preparation happens, who builds and books the outbound shipment into Amazon, how the plan responds when velocity changes in either direction, and how reserve stock stays visible to whoever makes the decision.
The failure mode is a warehouse that quietly becomes the place surplus inventory goes to rest. Buffer stock is only buffer stock if it has a defined role in the replenishment loop. Otherwise it is storage cost with a better name.
Do not forget the other channels
A seller entering Germany through Amazon frequently ends up running more than Amazon: an own store, eBay, other marketplaces, B2B orders, wholesale into retail.
If all of the inventory is committed to one marketplace fulfilment system, supporting an additional channel usually means creating a separate stock position, with its own freight, its own forecast and its own dead stock. An independent inventory layer keeps the same units available to whichever channel needs them.
This is a design consideration, not a rule. A seller with one channel and no plan to add another does not need it. A seller who expects to add one within the year should decide while the stock position is still theoretical.
Plan the reverse journey before the first outbound order
Two flows run backwards, and both need an address before they need a process.
Amazon removals. Inventory leaves FBA for several reasons, some the seller's choice and some not. When it does, it needs somewhere to go and something to happen to it: receipt, inspection, grading into sellable and non-sellable, relabelling, rework where the unit is recoverable, return to available stock, onward shipment to another channel, or disposal where recovery costs more than the unit is worth. A removal sent to an address with no process behind it becomes a pallet nobody counted.
Customer returns. For any D2C or multichannel volume, the same questions apply one unit at a time. Where the parcel goes, who opens and checks it, what decides restock versus rework versus write-off, and how a recovered unit re-enters available inventory rather than living permanently in a returns corner.
The return address and the recovery process should be settled before the first sale, not designed after the first return appears.
Before the first pallet moves
If these twelve have answers, the German operation is designed. If several do not, the freight booking is premature.
- Where the inventory is physically going first.
- Whether that destination is Amazon or an independent warehouse.
- Who is receiving the goods and who books the leg into them.
- What receiving information the warehouse needs, and by when.
- Whether the inventory requires preparation before it is usable, and where that happens.
- How much stock goes into Amazon immediately.
- Where the reserve stock sits.
- How replenishment into Amazon is triggered, sized and executed.
- Whether the stock has to support non-Amazon channels.
- Where Amazon removals will be delivered.
- Where customer returns will go.
- Who handles damaged, incorrectly labelled or unexpected inventory, and with what authority.
The milestone after the milestone
“German VAT number received” is a real milestone and clears a real obstacle. It is also closer to the starting line than the finish, because it answers a compliance question and leaves the operational ones open.
The next milestone is the one worth planning for: the inventory flow is designed before the first pallet leaves. Germany becomes operational when a seller knows not only where the stock is registered, but where it will land, who will handle it, what has to happen before it becomes sellable, how it will be replenished, and where it goes when the normal flow breaks. That plan is cheap to write in advance and expensive to retrofit around inventory that has already arrived.
The VAT side of the same decisions:
Planning a German warehouse? The registration is the slow part.
Stock in a German warehouse creates a German VAT registration obligation from the first unit, with no threshold, and the number takes four to eight weeks for an EU-established company and six to twelve from the UK or the US. That is longer than the freight, so it is worth starting when the warehouse contract is signed rather than when the pallets land. Vaytax handles the registration plus your monthly and annual returns, filed in-house by a licensed German tax advisor. €1,499/year all-in with registration included, or €99/month if you already have a German VAT number.
Start registration Check what you need first