What Happens to the Milk You Didn’t Pick
When fresher milk gets picked first, older cartons can be left behind to expire. New research shows how much food waste this creates at scale.
Picture the scene: you’re at the supermarket and two cartons of milk are on the shelf. One expires in three days, the other in five. You take the fresher one.
It’s a small, instinctive choice. But multiply it across thousands of shoppers and hundreds of products every single day, and it becomes a significant source of food waste that most grocery retailers have struggled to measure at scale.
Assistant Professor Varun Karamshetty from the Department of Information Systems and Analytics at NUS School of Computing is part of a research team that has produced the first large-scale evidence of how much grocery store food waste traces back to the sale of newer units when the older units remain on the shelf.
The study, published in Manufacturing & Service Operations Management (2026), was conducted with a large European grocery retailer with annual sales exceeding €10 billion, working across approximately 15.3 million sales transactions, 2,243 products, and 125 stores.
A Behaviour Nobody Was Counting
Grocery retailers track expiration waste carefully – food discarded because it crossed its expiration date unsold. What their data has never captured is how much of that waste is driven by the older units that stay behind because fresher units are sold instead.
Field studies cited in the paper found that in nearly 65% of observed purchase instances (22,537 out of 34,464), shelves displayed products with multiple expiration dates simultaneously. Shoppers reliably reached for the freshest option – a pattern the researchers term LEFO, or last-to-expire, first-out. The units left behind sit on the shelf until they expire and get discarded. The study calls this surplus “multiple-dates-led expiration waste,” or MDEW.
Standard point-of-sale systems record what was sold and what was thrown away, but not which specific units a shopper chose or passed over. MDEW doesn’t show up in conventional retail data at all.
Putting a Number on It
The team built a methodology using data that retailers already collect: daily sales, delivered quantities, closing inventory, and recorded expiration waste. By applying inventory flow equations across these variables, they computed lower and upper bounds on how much observed food waste could be attributed to the multiple-dates effect.
Expiration waste across the study averaged 23% of total sales at the product-store-week level. Of that waste, MDEW’s lower and upper bounds averaged 25% and 52% respectively. In a subsample covering 97% of the products studied, where the methodology yielded a precise figure, MDEW accounted for an average of 34% of expiration waste – roughly a third of food expiring in these stores.
The effect was strongest in fresh produce. Among perishables, MDEW accounted for an average of 50% of expiration waste, compared with 31% in bakery and protein deli, and 30% in ready-to-eat products. Franchise stores consistently showed lower MDEW than self-operated ones.
“Most retailers are working hard to reduce food waste. But without knowing which part of the process contributes how much waste, it is difficult to know where to start. That’s what this methodology gives them,” says Prof Karamshetty.
From Measurement to Action
The methodology requires no new technology and scales across entire store networks. The team also developed a diagnostic framework to help managers identify whether waste in a given product category is being driven by shelf rotation behaviour or by structural factors like over-ordering and shelf space allocation – a distinction that matters when deciding where to act first.
The longer-term picture involves a shift already underway. GS1, the global barcode standards body, has committed to transitioning major US retailers to 2D barcodes – capable of encoding expiration date information at the unit level – by 2027. That would make MDEW more directly measurable and open the door to more targeted interventions, including dynamic markdowns on near-expiry stock.
Many retailers have committed to halving food waste by 2030. This research gives them a clearer picture of one source that, until now, they had no way of measuring at scale.
Read the full paper here: Ashish Kabra, Nitish Jain, Varun Karamshetty, “Multiplicity in Product Expiration Dates and Food Waste in Grocery Retail Stores,” Manufacturing & Service Operations Management (2026). https://doi.org/10.1287/msom.2023.0607
