On the measurement of inventory record inaccuracies

Inventory records are often inaccurate, and this is known to be the source of major cost and service inefficiencies in retailing. In today's e-commerce and omni-channel environments, customers increasingly expect real-time visibility of stock availability across locations, making inventory record accuracy a prerequisite for reliable services such as in-store pickup or home delivery. However, the practices of measuring and reporting errors in inventory records vary considerably across the sector. Further, and while the advantages of error-free stock records are apparent, retailers may tolerate minor discrepancies as inconsequential, blurring the line between 'accurate' and 'inaccurate' records. We set out, for the first time, to: i) understand which inventory record inaccuracy (IRI) measures have been proposed so far in the literature, ii) identify how inventory record inaccuracy measurement takes place in retail practice and how it is organizationally embedded, and iii) propose a set of measures to enable benchmarking and continuous improvement. To do so, we first review the pertinent literature using a systematic search and selection method, followed by interviews with 25 retail executives responsible for areas such as inventory loss and prevention. Results obtained from the literature review and the interview study were validated in a workshop involving 46 retail professionals. We find that retailers employ a wide variety of IRI metrics - with simple binary measures being the most common ones - and that there is little consensus on what constitutes an acceptable measure. We propose a set of desirable attributes for the error measures and offer several insights that should be valuable for any retail professional involved in inventory decisions making.