Perpetual vs periodic inventory: which are you running?

A periodic inventory system does not track stock continuously. It establishes a figure by counting, at intervals — monthly, quarterly, annually — and derives the cost of what was sold in between by arithmetic: opening stock, plus purchases, minus closing stock. Between counts, the system does not claim to know what is on the shelf, and it is honest about that.

A perpetual inventory system updates the record on every movement. A goods receipt raises the on-hand quantity; a delivery lowers it. The system holds a running figure at all times and can tell you the cost of goods sold without waiting for a count.

Both are legitimate. Which one is appropriate depends on the value and volume of what you hold, and on whether decisions get made between counts. What is not legitimate is believing you run the second while actually running the first.

Here is the common shape. The software is perpetual: every purchase and every sale posts a movement, and the stock figure updates live. But nobody counts. Or counts happen once a year for the accountant, and any difference is written off in a single adjusting entry that nobody investigates.

That is a perpetual system with no feedback loop, and it degrades in a specific direction. Every source of loss that does not pass through the software — breakage, theft, samples handed out, units delivered short, a receipt entered for the quantity ordered rather than the quantity that turned up — accumulates in the gap between the recorded figure and the shelf. The recorded figure only ever drifts one way, and it drifts silently.

The result is worse than periodic, because the number is wrong and confident. A periodic system tells you it does not know. A neglected perpetual system tells you a figure, and people plan against it.

• Gross margin becomes unreliable. Cost of goods sold is derived from movements, so if the movements are incomplete, margin is overstated — and it is overstated most on the lines that lose the most stock, which are exactly the lines you would want to act on.
• Reorder points stop working. A low-stock alert fires against the recorded quantity. If the recorded quantity is above the real one, the alert fires late, and it fires latest on the fastest-moving items.
• Stock value on the balance sheet is overstated, by the cumulative value of everything that left without a record.
• Trust goes. Once the warehouse team has been sent to pick something the system said was there twice, they stop trusting the system and start checking the shelf first — which removes the only benefit perpetual inventory had.

Three questions, and they are more diagnostic than any feature list:

1. When did someone last physically count any part of your stock, and what happened to the difference? If the answer is “last year” and “it was written off”, you are running periodic inventory with extra steps.
2. When a purchase order arrives short, what does the person receiving it do? If they record the quantity from the order rather than the quantity in the van, your movements are documenting intentions rather than events.
3. Can you name the last three stock adjustments and why they were made? If adjustments are frequent and unexplained, the perpetual record is being manually corrected towards reality rather than reflecting it.

Whichever system you run, the cost assigned to a unit leaving stock has to come from somewhere, and units bought at different prices are not interchangeable for accounting purposes. IAS 2, the international accounting standard covering inventories, sets out how inventory is measured and which cost formulas may be used — specific identification where items are not ordinarily interchangeable, and otherwise first-in-first-out or weighted average cost, applied consistently.

The practical consequence for anyone choosing software: the cost formula is not a preference to be changed when margins look disappointing. It has to be applied consistently across similar inventories, and switching it is an accounting policy change rather than a settings edit. A system that lets a user flip between costing methods per report is offering a feature that should make you nervous rather than comfortable.

1. Get the opening count right. Every figure downstream is a running total of movements applied to an opening number. If that number is wrong, the error never washes out — it persists until someone recounts.
2. Receive against the delivery, not against the order. The goods receipt is the event that should raise stock, and it should record what arrived. This single habit removes the largest source of silent drift.
3. Cycle count instead of stock-taking. Counting a slice of the catalogue every week — weighted towards high-value and fast-moving lines — catches drift while it is still small and traceable, and it does not require closing for a day.
4. Investigate differences before adjusting. An adjustment with no explanation is a data-entry operation, not a control. The point of the count is the reason, not the correction.
5. Make returns reverse movements rather than delete records. A deleted line is history that cannot be audited; a reversing movement is history that can.

None of that requires particular software. It does require software that records movements as events rather than letting quantities be edited directly — which is the design decision underneath how Stockroot handles goods receipts, deliveries and returns.