One Stock Pool, Three Storefronts: Multichannel Ecommerce Fulfillment Without Overselling

Almost nobody runs one storefront for long. A store adds a second sales channel, a market place listing appears, and each of them wants to answer the same question at the same moment: is this still available? From the rack side, that answer is not a number in a column. It is a queue with a rule about who is allowed to sell the last unit, and the rule matters more than the sync interval. Ecommerce fulfillment services for a seller on three channels are judged on one thing: whether multichannel fulfillment answers every storefront from the same commitment instead of from that storefront's own reading of the shelf.

A stock figure is a snapshot; a channel sells against it as if it were a promise

The storefront that pulled its inventory in the morning is quoting a position that an afternoon dispatch has already changed. With three channels reading the same snapshot on three different timers, overselling is not a hazard you might encounter, it is arithmetic nobody finished. The units that got sold twice are the gap between the reads, and no integration setting closes a gap that is defined by timing.

Commitment is the field that carries the weight

A reservation rule says a paid order removes that unit from the pool visible to every other channel before a picker has touched it. Without commitment, the pool can go negative inside the interval between two pulls, and the recovery is a cancellation or a backorder message written in apology. Our bench works from committed quantities, and an order that reaches us with nothing reserved behind it lands in the exceptions list rather than in the pick path. The per-order decision page sets out how a line is routed once it is committed.

Three shapes of failure, all invisible until the carton does not exist

Pulls out of order produce phantoms: a return lands a unit back in the pool, one channel sees the increase and another still sees the old low figure, and both readings were true at their own moment. A cancellation that arrives after allocation releases a unit a packer has already boxed, so the count is right and the shelf is wrong. Bundles are the nastiest case, because sellable kits equal the minimum of their components; a channel that counts a kit against one line will happily sell kits the bench cannot assemble. Kitting is a stock question before it is a packing question, and our kitting and bundling page is where that constraint is written out.

A buffer is a commercial choice, not a safety default

Holding units back from the sellable pool buys tolerance for sync lag and costs turns. Too large and you own stock no customer can buy; too small and the buffer shows up as a cancellation. We will set one on request and will tell you how many days of cover it consumes, but the size is your decision, because the revenue that moves with it is yours.

What a clean setup looks like on our side of the wall

One source of truth for physical stock: the warehouse ledger, counted at goods-in and cycle-counted afterwards. One mapping table between your SKU values and the physical item, maintained at intake and versioned whenever a labeling job changes what is on the box. One dispatch feed that reports what physically left the building, reconciled against your store order numbers and never against the timestamp a file reached us. Stock held for the China to United States lane or moving through the small-parcel dropshipping channel follows the same rule; only the cut-off differs, and we publish that as 16:00 China time on the lines that run it.

Two things we do not do

We do not sell you a stock figure we have not counted, and we do not resolve a channel conflict by deciding which customer keeps their order. If the ledger says the last unit went out on a different channel, that is the answer we send back, with the dispatch record attached. Sellers who plan around that answer have a much easier quarter than sellers who plan around a number that flatters the position.

Questions this desk gets asked

Does sharing one stock pool across channels slow a shipment down?

No. The pool is a counting rule, not a speed limit; pick paths and the evening handover are unchanged, and only the sellable figure a channel sees moves.

Can one building hold two separate pools for two brands?

Yes, with separate bin locations and separate counts. The separation is charged as the extra handling it is, not as a free option.

What happens if a channel oversells anyway?

The unit either comes off the next delivery or the order is cancelled. We send the count that shows which it is, so you can decide before a customer finds out.

By Tony Zhang, FulfillDock by SBT. Written at the inventory-and-picking bench in Dongguan, and checked against this week's exception report before publishing. Published 23 October 2026, last checked 23 October 2026