New Research: When are inventory records accurate enough?


Previous research from ECR Retail Loss has identified that removing inaccuracies from stock records via a stock audit may lead to a substantial sales lift that has been found to be between 4% and 11%. Click here.

This finding has since been used by retailers to reposition counting as a “sales growth” strategy and introduce changes in count frequencies and a suite of interventions and new algorithms based off wrong inventory records.

However, the research to date has adopted a binary definition of accuracy, which in practice is not always the definition adopted by retailers. For example, retailers may decide to accept small errors in stock records for low-value SKUs sold in large quantities to reduce the effort associated with stock audits.

The extent to which, and how different accuracy definitions influence store performance, as compared to the binary measure of accuracy, has not been investigated yet. This is the purpose of this study.

Research objective and research questions

This study has two main objectives: First, it intends to investigate current practices regarding the measurement of inaccurate inventory records (IRI) in the retail sector. The literature discusses several different IRI measures, such as a binary representation of IRI, absolute IRI, an accuracy range or accuracy relative to stock holding, but it has not been studied so far which IRI measures retailers use in practice, and what the motivation for using a particular IRI measure is.

The study aims to close this research gap. Secondly, it aims to investigate the relationship between the traditional binary view of accuracy (i.e. accurate or not accurate) to the different definitions most widely adopted by the retailers.

Research questions to be investigated include:

Which IRI measures are currently used in the retail sector, and what motivates the use of these measures?

How do different IRI measures influence store performance indicators, such as sales, on-shelf availability and  counting strategies?

How is the influence of the IRI measures on store performance moderated by SKU and store characteristics?


The research project consists of two main phases.

Phase 1 is an exploratory phase that starts with interviews followed by a survey of the ECR Retail Loss IRI working group (circa 100+ retail business’s). The objective of Phase 1 is to establish current practices on the measurement of IRI, understand variations by retailer type, category, etc., link inventory record accuracy to possible different performance metrics, and to develop the specific research hypotheses to be investigated in Phase 2.

Phase 2 is then an empirical investigation of ideally six retailers to answer the research questions. This phase aims to cluster SKUs based on their IRI magnitude to understand which IRI measure performs best for which SKUs and which retail context and performance measure. To the extent the collected data permits such an analysis, results will be broken down in terms of industry, product category, and online vs. offline channels.

Research Team

The research project will be executed by three recognised experts in complementary fields:

Yacine Rekik: Professor in ESCP Business School (France) and expert in inventory control and inventory inaccuracies.

Aris Syntetos: Professor in Cardiff Business School, Cardiff University (UK) and expert in inventory forecasting and statistical theory.

Christoph Glock: Professor in Technical University of Darmstadt (Germany) and expert in inventory control and warehousing operations. 

Next Steps

The research will kick off in early 2024, and be complete by the end of 2024. If you would like to participate, please email Colin Peacock at

Feb 12, 2024