Counting stock without stopping the warehouse: scope, variances and decisions

Short answer: a stock count does not have to cover the whole warehouse at once. In NG WMS the scope is set by zone, location range, item, brand or LPN. Before release you see what falls into the count and which open documents touch it, and counting can be blind. Variances do not go straight to the ERP: each gets a decision on the worksheet, disputed locations are recounted, and the result is sent only after approval.

Why a full count gets in the way of everything else

When everything is counted at once, receiving and dispatch usually wait, because nobody wants stock to move while it is being counted. That makes sense once a year. But most stock problems are not in the whole warehouse; they sit in one zone, one item group or a handful of locations where mistakes keep happening.

So the first question is: what exactly needs counting.

Step 1: a scope that fits the problem

• The whole warehouse, with or without empty locations.
• Only locations holding stock, when empty locations are not in question.
• Targeted, by area, sub-area, zone, location, location range, item, brand, category, group, LPN or serial number — with include or exclude.
• From an Excel file: the file is checked first, and the import only goes through if every row is valid. Errors are shown by row number.

How a warehouse is divided into areas, sub-areas and locations is described on the Warehouse model page. The more precise the model, the narrower a targeted count can be, and the quicker it is to prepare.

Step 2: preview and pre-check before release

Before tasks are created you see what falls into the count: locations holding stock, empty locations and those where confirming an empty location is mandatory. Scope mistakes are caught here, while fixing them is still simple.

The second part is the pre-check of documents still open within the scope. If receiving or picking is under way at those locations, counting would catch stock on the move. In strict mode the count is not released while such documents exist; in tolerant mode it is released, but the document list stays visible.

Step 3: counting

Tasks are assigned per location, with attempts and a sequence, and progress is tracked as counted locations out of the total. Counting is done in the scanner app. With blind counting the counter does not see the expected quantity, so what they record is genuinely what they counted. You also decide whether a new LPN or an item outside the scope may be added.

Step 4: the variance worksheet

Each variance gets a decision — accept, recount, ignore or set to zero — per line or in bulk, always with a reason. A variance that affects stock already reserved is not applied automatically and requires an explicit decision.

Step 5: recounts and cross-zone analysis

An example: zone A is short by 12 units of an item, while zone C holds 12 units too many of the same item. The net variance for the item is zero. That is not a loss but stock put away in the wrong place. Cross-zone analysis pairs such cases automatically, and a recount confirms the finding before the decision.

An EAN scan anomaly report also singles out suspicious reads, and the count audit keeps cancelled attempts.

Step 6: approval, then the ERP

The count goes through review and, where needed, a recount. It is then approved in the WMS, sent to the ERP and closed once the ERP confirms receipt. Posting the variances happens on the ERP side; the WMS sends it a result that has already been checked. The full flow is described on the Cycle count page, and the exchange with the ERP on the ERP integration page.