Skip to main content

Warehouse · 3 min read

Multi-location stock control for FBA sellers

FBA sellers with prep centers, warehouses, and Amazon stock need one reservation model. Without it, inbound plans overpromise units.

By Kenderson Tripaldi · May 5, 2026

Operator checking inventory levels on a tablet beside warehouse shelves

Once inventory lives in more than one place, a simple on-hand number becomes dangerous. A seller may have units at a prep center, in a 3PL warehouse, reserved for an inbound plan, in transit to Amazon, available in FBA, and pending removal. Counting all of them as available creates oversold replenishment plans.

The fix is a reservation model.

Separate physical stock from available stock

Physical stock answers "where are the units?" Available stock answers "what can we commit?" A unit can be physically in your warehouse but unavailable because it is reserved for a shipment, under quality hold, or allocated to a wholesale order.

Track these states:

  • on hand by location
  • reserved by plan or order
  • in pick
  • packed
  • in transit
  • received by Amazon
  • unavailable or hold

Reservations prevent double counting

When an inbound plan is created, reserve the units immediately. The warehouse can still see the physical stock, but other workflows cannot consume it. If the plan is cancelled, release the reservation. If the plan is partially packed, reserve only the committed quantity and return the rest to available.

Reconcile location movements

Every move should have a reason: receive, transfer, pack, adjust, cycle count, removal receive, or plan cancellation return. The reason matters during audit. A negative adjustment means something very different from a transfer to Amazon.

Multi-location control is less about fancy warehouse software and more about refusing to let "on hand" mean five different things. Once availability is explicit, inbound planning becomes calmer and much harder to overpromise.

Add cycle counts where decisions happen

Cycle counting should focus on locations that drive commitments: pick bins, staging lanes, packed-but-not-shipped areas, and removal receiving. Counting only deep storage can miss the places where availability changes fastest. If a shipment plan reserves units from a pickable location, that location should be counted often enough to trust the reservation.

Use count results to classify errors. A missing unit in pick may indicate a packout scan issue. A mismatch in staging may indicate units were moved without a transfer. A removal receiving mismatch may create a reimbursement case. The location of the variance tells the team which process needs repair.

Keep reservations visible to operators

Reservations should not be hidden accounting fields. Warehouse operators need to see why stock is unavailable. A unit reserved for an inbound plan should look different from a unit on quality hold or a unit allocated to another sales channel. When the reason is visible, operators are less likely to work around the system.

Visibility also helps managers make tradeoffs. If a high-margin order needs stock reserved for a slow inbound plan, the team can intentionally release or reassign the reservation. The control is not meant to freeze the business. It is meant to make every commitment explicit.

Audit negative availability

Negative availability is a warning that the model allowed more commitments than the operation could support. Review every negative state by source: reservation timing, late scan, manual adjustment, transfer delay, or cycle count correction. The cause matters because the fix differs. A reservation timing problem needs workflow sequencing. A cycle count correction needs physical inventory control.

Do not simply reset the number to zero. That hides the failure. Record the adjustment reason, affected plan or order, and owner. Over time, negative availability should become rare. If it does not, the team is planning from numbers it cannot trust. That trust is the foundation for every inbound plan, pick wave, and channel allocation decision. When trust drops, freeze new commitments until the affected location is recounted and reconciled. That pause is cheaper than building plans from bad stock data.

Keep reading

View all posts