Inventory management software: position, policy and traceability

The stock position and its many states

A usable position distinguishes physical on-hand from what may be sold or consumed. Quantities sit in states: available, reserved against demand, in quality inspection, blocked, in transit between sites, on order from a supplier, and returned pending disposition. Reporting a single number without those states is what produces stock that exists on a screen and not on a shelf. Where goods are held under a customs or excise procedure, the record also has to identify their status, because movements in and out of duty suspension are reportable to the supervising administration.

Policy: what the system decides on your behalf

Replenishment configuration sets how much cover to hold and when to reorder, typically through a reorder point with a quantity, a periodic review to a target level, or a min-max band. Each parameter encodes assumptions about demand variability, lead time and service ambition, and the parameters go stale as those change. The practical failure is not choosing the wrong method but never revisiting the settings, so items keep ordering to a pattern that suited them years ago. Systems that expose which parameters have not been reviewed are more useful than those offering more sophisticated methods.

Traceability by batch, serial and expiry

Regulated, perishable and safety-critical goods need to be traceable from receipt to despatch, which means the identifier travels through every transaction rather than being recorded only at goods-in. That supports rotation rules such as first-expiring-first-out, targeted recall by batch rather than by product, and evidence for quality investigations. Retro-fitting traceability is disproportionately painful, because historical stock has no identifier and physical units must be re-labelled, so the requirement should be settled before implementation rather than added when a customer asks.

Counting and the tolerance for being wrong

Record accuracy decays through mis-picks, unrecorded damage, transaction errors and theft. Periodic full counts halt operations and find errors late; cycle counting samples continuously by value or movement class and finds them sooner. Whichever is used, the important control is what happens after a discrepancy: adjustments should require a reason code and an approval above a threshold, and repeated discrepancies in the same area should trigger investigation rather than another adjustment. Accuracy targets differ by business, but an untracked adjustment rate always understates the problem.

Boundary with warehouse execution and with planning

Warehouse systems own location-level detail and the work that moves stock; inventory systems own the aggregate position, its states and its policy. Network-level questions, how many stocking points, where to position cover, how to shape the network, belong to supply chain planning and are a different discipline again. Drawing those lines explicitly avoids the common architecture where three systems each hold a partial truth and a spreadsheet reconciles them monthly.

Frequently asked questions

Can the warehouse system be the only inventory system?

For a single site it often can, provided finance and purchasing consume its data rather than keeping parallel records. Multi-site operations usually need an aggregate layer, since each warehouse system knows only its own building.

How are consignment and customer-owned goods handled?

Through an ownership attribute on the stock record, so the same physical location can hold goods belonging to different parties. Without it, contract warehouses cannot bill or report accurately and valuation includes stock the business does not own.

What signals that inventory data has stopped being trusted?

Planners keeping side spreadsheets, buyers ordering against their own estimates, and customer service checking with the warehouse before promising. Those behaviours appear well before accuracy metrics deteriorate visibly.