Amazon's inbound routing algorithm is reasonably well-designed for a business with one warehouse. You create a shipment, Amazon tells you which fulfillment center to send it to, you send it there. The system is trying to distribute inventory across its network so that customers everywhere get fast delivery. For most sellers, following that recommendation is fine.
Our Place is not most sellers. And for Our Place, following Amazon's routing recommendation turned out to be consistently wrong.
The setup
Our Place runs 7 warehouses globally: California (CAO2 and CAO3), Texas, Pennsylvania, Illinois, Ontario in Canada, and locations in the UK and Australia. That is a lot of surface area for a routing decision. When you create an FBA shipment, Amazon's system looks at where it wants the inventory to go and recommends a source location based on its own logic — logic that optimizes for Amazon's network distribution goals, not for the seller's freight cost or check-in timeline.
For brands with one warehouse, Amazon's recommendation and the seller's optimal choice are usually the same because there is no choice to make. For a brand with seven warehouses spread across the US and two international markets, Amazon's recommendation is just one option among several, and it is rarely the best one.
What Amazon kept recommending
Amazon's algorithm frequently directed Our Place to send inventory from Texas, Pennsylvania, or Illinois. From Amazon's perspective, those locations might be closer to the FCs it was trying to populate for geographic distribution. The problem is that "closer to Amazon's preferred FC" is not the same as "best for the seller."
From a cost standpoint, shipping large oversized items like Wonder Ovens and Dream Cookers from a Midwest or East Coast warehouse to the FCs Amazon designated often meant longer freight distances, higher per-unit costs, and slower transit times. California warehouses, by contrast, had established freight lanes into West Coast FCs with faster check-in times.
From a supply limit standpoint, the California warehouses (CAO2/CAO3) were where the team had the strongest existing inbound history and the best utilization relationship with those FCs. That history matters because Amazon's supply limit calculations are partly informed by your inbound track record at specific FCs.
The California-first rule
The rule the team established was simple: for US FBA shipments, always source from California warehouses first. Do not follow Amazon's routing recommendation to Pennsylvania or Illinois or Texas. Override it.
This is a manual decision. Amazon does not automatically route based on your freight cost optimization — it routes based on its own network optimization. If you want to optimize for your own operation, you have to make that choice explicitly when you create the shipment.
The California-first rule produced measurable improvements on two dimensions. FBA check-in times were faster, because the California-to-West-Coast-FC lane was well-established and high-volume, which means Amazon processed those inbound shipments quickly. Freight cost per unit was lower, because the lanes out of CAO2 and CAO3 were more competitive than equivalent lanes from Midwest facilities for the oversized category.
The third benefit was less obvious: supply limit headroom. Because the California warehouses had the strongest inbound history with the relevant FCs, sends from those locations were less likely to trigger "high supply" flags that would block shipment creation. Amazon's supply limit system is not uniform across all source-destination pairs. Routing from the warehouses with established history at the destination FCs kept that history active and the limits open.
Why Amazon's algorithm fails for brands like this
Amazon's routing recommendation is doing something rational from its own perspective. It is trying to pre-position inventory across its fulfillment network to minimize its own last-mile delivery costs. That goal is sometimes aligned with the seller's goals and sometimes not.
For a brand with one source location, there is no conflict. But for Our Place, with seven warehouses and specific constraints around oversized items, freight costs, supply limits, and check-in speed, Amazon's recommendation is solving the wrong problem. It is optimizing Amazon's distribution, not the seller's operation.
The other thing worth understanding is that Amazon's routing recommendation carries no authority. You are not required to follow it. It is a suggestion built on Amazon's data, not yours. Your data — actual freight rates by lane, actual check-in times by FC, actual supply limit history by source-destination pair — will produce a better decision for your business than Amazon's algorithm will.
Build your own routing logic. For most brands with multiple warehouse locations, the answer will involve more California than Amazon recommends.
