Inventory record inaccuracy (IRI) is the mismatch between the quantity that is recorded in a company's inventory management system and the quantity that is actually physically available. IRI can lead to significant issues in retail, e.g., by causing stockouts and revenue losses triggered by unnecessary replenishment.

This paper evaluates the effects of IRI on retail store inventory and sales management performance. We propose a novel network data envelopment analysis (NDEA) model, capable of setting store-level performance standards more accurately than state-of-the-art models. To support managers in identifying the root causes of IRI and in setting realistic target for mitigating IRI, the insights of the proposed NDEA model are used to develop two novel performance indicators: the IRI improvement potential and the IRI improvement workload.

This research uses real-life data of an international fashion retailer. The data set contains information of more than 5,250,000 inventory items kept in 81 retail stores. The computational experiments show the benefit of using relative measures to quantify IRI levels accurately across SKUs. Furthermore, decomposing store-level management into inventory management and sales management is found to be highly beneficial for evaluating the impact of IRI on store-level performance. Numerical results also demonstrate that IRI improvement is small for near-efficient stores and remarkably large for highly inefficient stores.

库存记录不准确性（IRI）是公司库存管理系统中记录的数量与实际可用的数量之间的不匹配。IRI可能导致零售中的重大问题，例如，由于不必要的补给而导致缺货和收入损失。

本文评估了IRI对零售商店库存和销售管理绩效的影响。我们提出了一种新颖的网络数据包络分析（NDEA）模型，该模型能够比最先进的模型更准确地设置商店级别的性能标准。为了支持管理人员确定IRI的根本原因并确定减轻IRI的现实目标，提议的NDEA模型的见解被用于开发两个新颖的绩效指标：IRI改进潜力和IRI改进工作量。

这项研究使用了国际时装零售商的真实数据。该数据集包含81家零售店中保存的超过525万库存物品的信息。计算实验表明，使用相对措施来准确量化整个SKU中的IRI水平是有好处的。此外，发现将商店级别的管理分解为库存管理和销售管理对于评估IRI对商店级别的绩效的影响非常有益。数值结果还表明，对于效率高的商店，IRI的改善很小，而对于效率低的商店，IRI的改善却很大。