Inventory

Owning four thousand units is not the same as having four thousand to sell

A total inventory number answers a question about ownership. Planning needs the answer to a different one: how many of those units could go in a box for a customer this afternoon. Between the two sit the reserved buckets, the inbound pipeline and the damaged units going nowhere. The SKUs and quantities below are invented so the arithmetic can be checked.

Total inventory is a claim about ownership. It says these units exist and they are yours, wherever they are sitting. It says nothing about whether one of them can be picked, packed and shipped to a buyer today. Some are already spoken for. Some are on a truck between two warehouses that belong to the same fulfilment network. Some have not been received into the network at all. Some are in a damaged-goods pile waiting for a decision.

Three words get used as though they were one, so this article uses each of them a single way. Total inventory, which this article also calls owned, is every unit that is yours, received or not. On hand is the part that has been physically received into the fulfilment network — check what your own source system puts in that column, because that is the definition your numbers will follow. Available is the part of on hand that can be sold today.

The number you can sell today is a strict subset of the number you own, and the distance between them moves with your order volume, your inbound schedule and how much stock is in motion. Plan against the wrong one and you can place purchase orders late while a report tells you everything is fine.

The report was not lying to you. It was answering an accounting question while you were asking an operations question.

Step 1 — Break the one number into four

Take one SKU in one fulfilment network. Call it SKU A, a kitchen scale, with 4,000 units to its name. That figure resolves into four states that behave completely differently:

In SKU Compass they sit behind a stock cell in the FBA Stock Breakdown, and each row is labelled with what it can do: Available is described as sellable right now, Inbound as on the way and not yet sellable, Reserved as held and not sellable, and Unfulfillable as damaged or otherwise unsellable.

Three of those four states are on hand: available, reserved and unfulfillable add up to 3,400 units physically received into the network. The 600 inbound are owned but not yet received, which is why the 4,000 is a total rather than an on-hand figure.

Say SKU A sells 48 units a day — invented, like everything here, but hold it still. Divide each state by it and the rows stop being an inventory picture and start being a calendar. 4,000 owned ÷ 48 = 83.3 days. 2,680 available ÷ 48 = 55.8 days. The difference is 1,320 units, or 27.5 days of cover that exists on the report and not in the warehouse.

Step 2 — Reserved is three situations wearing one label

For today’s sellable quantity, reserved is unavailable. For replenishment planning, the sub-buckets matter, because some reserved units may return and some will not. The units sitting under that one label are not doing the same thing.

The breakdown splits reserved into sub-rows labelled Customer orders, FC transfer and FC processing, and it splits inbound the same way, into Working (prepping at you), In transit (shipped) and Receiving (at FC). Those sub-rows are the whole point. They turn one unusable subtraction into three decisions.

SKU A’s 640 reserved units, split:

Ask one question of each sub-bucket: are these units expected back, or are they already committed to demand you have taken? The 410 held against customer orders are committed, and counting them as cover counts the same units twice, once as a sale and once as stock. The 160 in transfer and the 70 in processing are still yours and still inside the network, in a delay rather than a departure. Treat those as potentially available later, once the transfer or processing completes, rather than as lost. A bucket name is a reason to expect units back, not a date they will be back on.

So of 640 reserved units, 410 are already committed to demand you have taken and 230 are pending. That single split moves the number you can plan against from 2,680 to 2,910 without a unit arriving, for as long as the 230 stay marked as expected rather than as banked.

Inbound rewards the same scepticism. Units in transit have physically left you; units marked working are still on your own floor. One inbound figure hides which you can count on soonest.

Step 3 — Four numbers, and what each one is for

Once the buckets are split, SKU A does not have one stock level. It has four, and each answers a different question. All of them are correct. The days of inventory formula is the same division in every row; only the numerator changes. Using the wrong numerator is what goes wrong:

Read the top and bottom rows against each other. The owned figure is 490 units above the pipeline figure, and those 490 are the 410 already sold plus the 80 unfulfillable. Neither group is going to ship to a new customer. That is 10.2 days of cover that cannot happen, sitting inside a number a planner will treat as stock.

Watch the decision flip

Give SKU A a reorder buffer — lead time, shipping time and safety days added together — of 60 days. That number is yours to set from your own supplier history, not one anyone can hand you. Sixty is what this example uses so the comparison has something to bite on.

Step 4 — Two SKUs, the same owned number, opposite problems

That gap is easy to dismiss as one unlucky SKU. The sharper version is two SKUs reporting the same owned figure. Here are SKU B and SKU C, each with 1,800 units owned, each selling 30 a day:

On the owned line they are identical twins: 1,800 ÷ 30 = 60.0 days each. Split the buckets and they stop resembling each other at all:

At the sellable line they are 24.0 days apart. By the bottom row they are the same number again, which is exactly how a planner ends up treating two different situations identically.

SKU B has a supply problem. It is at 53.7 days against the 60-day buffer, and its pipeline row is the same as its reserved row, because nothing is on the way. Whatever returns from transfer and processing is 30 units, a single day. B needs a purchase order, and the owned figure of 60.0 days was the only thing hiding that.

SKU C has a timing problem. It is at 29.7 days of sellable cover, which is worse this week. But 500 units are inbound and 275 more are expected back from transfer and processing. On these numbers, C’s immediate problem is timing rather than an obvious need for another PO: it needs the 500 to arrive inside 29.7 days. That is a question about a date, answered first by chasing a shipment rather than by raising a purchase order.

Step 5 — Plan off the number that can reach zero

The rule falls out of the arithmetic: start with available stock, then add only future supply you have a concrete reason and a date to trust. That is the quantity that can run down to nothing and leave an order unshipped, and it is the one to plan replenishment against.

SKU Compass makes the equivalent choice explicit one level further out, at the purchase-order line. A Counting stock setting decides whether days of stock counts units already ordered on open purchase orders, and its help text spells out the trade: including them assumes those orders arrive as expected, while the stricter on-hand-only setting tells you to order sooner. It states the consequence too — the same SKU can look comfortable on one setting and urgent on the other. The measure it feeds is labelled Days of Supply.

Which gives you a test rather than a threshold. Run a SKU on both bases. If the answers land on the same side of your buffer, the basis is deciding nothing. If they land on opposite sides — the way SKU A does at 55.8 and 73.1 days against a 60-day buffer — that SKU’s plan is being decided by your choice of denominator rather than by its stock, so make the choice on purpose.

What will bite you anyway

Do not subtract the same units twice

Before you subtract customer-order reserved from a sellable figure, check what that figure already excludes. Some systems report an available quantity with open-order allocations removed already. Subtract the customer-order bucket from that and those units come out twice, which understates cover and triggers purchase orders you do not need. Learn what your starting number means, then adjust once.

Unfulfillable is owned, unsellable and does not fix itself

SKU A’s 80 unfulfillable units are 1.7 days of cover at 48 a day, and they will still be 80 units next month unless somebody removes them, reconditions them or writes them off. Treat that line as a task queue rather than as inventory and audit it on a cadence you set, unless you have checked that your own workflow dispositions that stock without being asked.

The buckets are a photograph, not a reconciliation

Each bucket is a state a unit passes through, so the split you read is true of the moment it was captured rather than of the day. SKU Compass says so on the panel: its footer note states that the breakdown is pulled from an hourly inventory snapshot, that the stock cell updates on a slower cadence, and that the two can differ by a few units between refreshes. Use a split to size a gap, not to reconcile to the unit against a number captured at a different second.

The method, in four lines

1. Start from owned, and stop calling it stock.
2. Remove what cannot ship: unfulfillable, and anything already allocated to a buyer.
3. Add back only what you have a concrete reason to expect, and label it with when — transfer and processing shortly, inbound on its own schedule.
4. Divide each figure by sales per day, compare the answers to your reorder buffer, and where the bases disagree, investigate the assumptions before acting.

That method moved SKU A from 83.3 days of imagined cover to 55.8 days of real cover, turning a decision to do nothing into an order already 4.2 days past its trigger. No new data arrived. The units had been sitting in those buckets the whole time.