Amazon Mexico Inventory Placement: NARF Versus Local FBA, Cost Per Unit

Selling on Amazon Mexico from a US base offers two placements and they are not variations of the same thing. Remote fulfillment ships each order from US inventory and pays a border crossing per order. Local FBA pays one crossing to get stock into the country and then pays storage until the unit sells. The choice is an arithmetic problem, and the crossover point moves with velocity.

Most sellers pick a placement once and keep it. The better approach is to place per SKU, because the same catalog can contain products on both sides of the crossover.

The two cost shapes

Remote fulfillment carries a per order cost that includes the crossing. Local FBA carries a one time inbound cost and a recurring holding cost. One scales with orders, the other scales with time.

The storage clock starts earlier than sellers expect

Amazon Mexico charges storage per cubic decimeter per month, at 0.36 MXN from January to September and 0.53 MXN from October to December. The seasonal step is roughly 47 percent, and it lands in the quarter when most sellers are holding their largest inventory.

The long term storage penalty begins at day 181, not at day 365. A unit that arrives in March and has not sold by September is already in penalty territory. Placement models built on a 12 month assumption understate the holding cost of slow movers by 6 months.

Holding stock on both sides at once changes the arithmetic again, and that case is worked through in Binational Inventory: What One Dashboard Across Two Warehouses Has to Reconcile.

Where the crossover sits

The comparison reduces to a single question per SKU: does the unit sell fast enough that its share of a single crossing is cheaper than the per order crossing it would otherwise pay.

Take a consolidation of 500 units. The fixed customs charges on that entry are the DTA floor of 258.91 pesos and a mid range brokerage of roughly 9,000 MXN, which is about 18.5 pesos per unit. If the unit sells within 60 days, it pays that 18.5 pesos plus 2 months of storage. If it sells within 300 days, it pays the same 18.5 pesos plus 10 months of storage plus a long term penalty that began at day 181.

The pattern is that the customs cost per unit falls as consolidation grows, while the storage cost rises with time. Fast movers should sit locally. Slow movers should not, and the boundary is closer to 180 days than to a year because of where the penalty starts.

The subscription fee changed the small seller arithmetic

Amazon Mexico restructured its subscription pricing, and the announcement is specific about the threshold:

"Si las ventas mensuales son menores a $26,000 MXN, la comisión de suscripción será de solo $75 MXN al mes"

Amazon México, announcement of 17 February 2026

For a seller below that monthly sales level the fixed platform cost is small enough that placement decisions are driven almost entirely by logistics rather than by platform overhead. Above it, the fixed cost is a constant in both scenarios and cancels out of the comparison. Either way it should be removed from the placement model rather than blended into a per unit figure.

Labeling decides whether local stock is sellable at all

A unit received into a Mexican facility without compliant Spanish labeling is not inventory, it is a problem waiting for a receiving report. The requirement is set in law:

"Los datos que ostenten los productos o sus etiquetas, envases y empaques y la publicidad respectiva, tanto de manufactura nacional como de procedencia extranjera, se expresarán en idioma español y su precio en moneda nacional."

Ley Federal de Protección al Consumidor

This makes the NOM file a placement input, not a compliance afterthought. A SKU whose label file is complete can be consolidated and sent. One whose file is incomplete should stay on remote fulfillment until it is fixed, because the alternative is paying to move stock into a warehouse that cannot release it.

What the placement decision needs as inputs

• Units sold per month per SKU, over at least 90 days rather than a peak month.
• Cubic decimeters per unit, because storage is charged on volume rather than on value.
• NOM label file status, as a yes or no gate before any other calculation runs.
• Return rate, because returns land locally and a local return is recoverable while a remote one usually is not.
• Seasonality, because the storage rate steps up 47 percent for the last quarter of the year.

The mistake that shows up in the fourth quarter

Sellers consolidate ahead of peak season, which is correct, and they size the shipment to peak demand, which is also correct. What they miss is that unsold peak inventory then sits through the highest storage rate of the year and walks into the day 181 penalty in the following spring.

The fix is not to send less. It is to plan the exit before the entry: decide in advance what happens to peak leftovers, whether that is a discount channel, a return north, or local disposal. A unit with no exit plan defaults to the most expensive one.

Removals are a customs event, and that changes the exit plan

Sellers treat an Amazon removal as a logistics chore. Across a border it is not. Stock sitting in a Mexican fulfillment center has already been imported, and moving it back north is a fresh import into the United States with its own declaration, its own evidence requirements and its own timeline.

The customs law frames clearance as a set of formalities in both directions, which is the point sellers miss when they plan an exit:

"Para efectos de esta Ley, se entiende por despacho aduanero el conjunto de actos y formalidades relativos a la entrada de mercancías al territorio nacional y a su salida del mismo."

Ley Aduanera, Article 35

Two consequences follow. A removal cannot be scheduled as a same week action, because the paperwork has a queue. And the fixed charges apply again: the DTA floor of 258.91 pesos and a brokerage minimum of 3,500 MXN are per entry, so removing 40 units costs nearly what removing 400 costs.

Why the exit has to be decided before the entry

A seller who consolidates 500 units in September and sells 380 by February has 120 units and 3 options. Discount locally, which recovers cash and keeps the units out of penalty. Consolidate a northbound removal, which costs a full set of fixed charges for 120 units. Or dispose locally, which recovers nothing but stops the storage clock.

All 3 are reasonable. What is not reasonable is discovering the choice in March, when the units have been in long term storage since day 181 and the peak season storage rate of 0.53 MXN per dm3 has already been paid across the fourth quarter.

A worked comparison on 120 leftover units

On low value units the middle row rarely wins, because roughly 3,759 pesos of fixed cost spread over 120 units is about 31 pesos a unit before freight. On higher value units it frequently does. The threshold is not a rule of thumb, it is unit value against roughly 31 pesos plus the northbound freight share.

Joint liability makes classification a placement input too

Sending stock into a Mexican fulfillment center means filing an entry with a tariff classification, and errors in that classification are not treated as clerical:

"Both importers and brokers will be liable for instances of undervaluation, tariff misclassifications, and false or incomplete customs entries."

Benesch, Friedlander, Coplan and Aronoff LLP

For placement this matters in a specific way. A SKU whose classification is uncertain carries more risk when it is imported in bulk than when it ships in small remote fulfillment consignments, because the bulk entry concentrates the exposure into one declaration. Resolving the tariff code is therefore part of qualifying a SKU for local placement, alongside the label file and the velocity check.

The three gates before a SKU moves local

• Label gate. The NOM file is complete and reproducible, so a damaged label can be replaced at the warehouse.
• Classification gate. The 10 digit code is settled, with the first 8 digits as the fracción arancelaria and the last 2 as the NICO.
• Velocity gate. The 90 day sell through implies an exit well before day 181.

A SKU that fails any of the 3 stays on remote fulfillment. That is not a lesser option, it is the correct option for inventory that is not yet ready to be capital sitting in another country.

What a placement review looks like in practice

Run it monthly against the previous 90 days rather than the previous month, because a single promotion distorts a 30 day window badly enough to move a SKU across the boundary and back again. Four columns are enough: units sold, days of cover at current stock, days since receipt, and label file status.

The SKUs that need action are the ones where days since receipt is approaching 181 while days of cover is still high. That combination means the unit is heading into penalty storage with inventory left, and it is the only pattern in the table that gets more expensive the longer it is ignored.

Everything else can wait a month. A fast mover with 20 days of cover does not need a decision, and a SKU that arrived 3 weeks ago has not yet produced enough signal to act on.

The prep step that sits in front of local placement is covered in Laredo Before the Crossing: Storage, Prep and the Work That Belongs North of the Border.

How BringGo Ship Handles This

BringGo Ship runs the three gates before any SKU moves to local placement: a complete and reproducible label file, a settled ten digit tariff code, and a ninety day sell through that clears the stock well before the long term storage boundary. Our Laredo facility prepares and consolidates the inbound, and our team plans the exit for peak leftovers before the entry rather than after it, so unsold seasonal stock does not walk into penalty storage in the spring.

Frequently asked questions

Is remote fulfillment always slower? It always includes a crossing, and a crossing is measured in days rather than hours. Whether that matters depends on the promise shown to the buyer before checkout.

Can I split a SKU across both placements? Yes, and for high velocity items with volatile demand it is often correct. The local pile serves the base rate and the remote pile absorbs the spikes.

Does the long term penalty really start at 181 days? The long term storage charge begins at day 181, which is why a 6 month sell through is the practical planning boundary rather than a 12 month one.