Researchers at Cardiff University have helped develop a new approach to inventory management that could help retailers improve product availability, reduce lost sales, and make stock counting more efficient.
The new report, Smart Inventory Record Inaccuracy (IRI) Prediction and Management , has found that many stock record errors are predictable and can be identified before they have a significant impact on customers and businesses.
The international study, from ECR Retail Loss, was co-authored by Professor Aris Syntetos from Cardiff Business School, alongside Professor Yacine Rekik of Emlyon Business School in France, and Professor Christoph H. Glock of University of Darmstadt in Germany.
Drawing on more than 1.3 million stock-audit observations from six major grocery retailers between 2018 and 2022, the research examined how stock inaccuracies arise and how they can be managed more effectively.
Inventory record inaccuracies are a longstanding challenge for retailers. In many cases, stock systems indicate that products are available when shelves are actually empty, creating what is known as 'phantom inventory'. These errors can lead to missed sales opportunities, frustrated customers, and inefficient replenishment decisions.
The research team developed a transparent machine-learning model that uses data already routinely collected by retailers to predict where inventory records are most likely to be wrong. The model can prioritise which products should be checked first, enabling businesses to focus their efforts on items most at risk of inaccuracies rather than carrying out blanket stock counts.
The study found that the approach correctly identifies genuine inventory record errors around nine times out of ten and detects approximately 19% more errors than leading conventional methods. It also flags more than 80% of phantom inventory cases before retailers experience significant sales losses.
Professor Aris Syntetos, Professor of Operational Research and Management Science at Cardiff Business School, said: "Retailers have traditionally tackled inventory inaccuracies by increasing the frequency of stock counts. Our findings show that a more targeted approach can deliver far better results. By using data that retailers already possess, organisations can focus attention where it is needed most, improving product availability for customers while making much more effective use of staff time and resources."
"This research demonstrates how advances in analytics and machine learning can support practical decision-making in complex retail environments. The goal is not to eliminate stock counting, but to make it smarter, more efficient and more impactful."
While statutory stocktakes will remain an essential part of retail operations, the researchers argue that predictive approaches can help businesses direct resources towards the products and categories where they will have the greatest impact. The report suggests that, in an increasingly complex retail environment, the future of inventory accuracy may depend less on counting more, and more on knowing exactly what to count.
The findings have attracted attention from retail professionals involved in inventory management and loss prevention. Colin Peacock, Group Strategic Coordinator at ECR Retail Loss, said: "This research shows us a smarter approach to stock records. Retailers can point their team at the items most likely to be wrong and find more problems in less time. So, they can catch the empty shelves before they cost sales."
Dennis Gibson, Shortage Control Manager at US retailer Staples, said: "The future of inventory accuracy isn't counting more, it's counting smarter. This research provides compelling evidence that predictive analytics can help retailers prioritise effort, improve availability, and reduce the hidden costs of inventory inaccuracy."
Abigail Lynch, Project Manager at ALDI UK and Ireland, highlighted the wider operational benefits: "As a discount retailer, balancing the labour we invest with maintaining an accurate stock file is key to keeping overheads low and ensuring customers can access high-quality products at affordable prices. Accurate stock records are critical in helping us minimise food waste, enabling us to operate more efficiently and responsibly. We congratulate the research team and ECR Retail Loss on conducting such a valuable study."
Read the report, Smart Inventory Record Inaccuracy (IRI) Prediction and Management, in full here .
