The positions to be checked during a sampling inventory consist of different subsets (recording scope). These include, on the one hand, particularly high-value items that the company wishes to verify out of its own interest (= full count positions) and, on the other hand, the calculated sampling positions.
In most estimation and test procedures allowed for inventory sampling, the sample size has a significant impact on the outcome. The following rule of thumb applies: the more homogeneous the values of the individual population elements (articles, bins, item subsets, etc.) are, or the smaller the range between the lowest and highest values, the smaller the required sample size will be.
Legally, a maximum sampling error of 1% is allowed when applying an estimation method. Sampling is considered „efficient“ if this upper limit is not exceeded while also achieving a minimal recording scope. It is common practice, however, to include a buffer in the scope to account for the possible effects of deviations.
When defining the calculation parameters for implementing inventory sampling, it is therefore useful to consider past results as well as an assessment of the current inventory accuracy. This is the only way to reduce effort to a minimum. At present, JC is testing an AI-based approach that does not rely solely on a warehouse’s historical data, but also incorporates data from thousands of previously conducted sampling inventories.