Retail Insight launches AI waste prediction for grocers

Retail Insight has launched predictive artificial intelligence features on its WasteInsight platform for grocery retailers managing food waste.

Earlier warnings

The new feature, Predictive Waste, uses delivery schedules, item perishability data, store-level demand patterns, and customer basket data to identify stock at risk of not selling before expiry. The system can alert retailers as soon as sales and inventory signals show an item may not sell through in time.

The earlier warning is designed to help store teams apply discounts several days sooner than in traditional waste processes. Staff can receive predictive markdown prompts up to a week before expiration, rather than relying on last-minute checks close to the final day a product can be sold.

Food waste remains a major cost for the retail sector. Retail Insight cited research covering 3,500 retailers that estimated the annual cost at more than GBP £400 billion, making it one of the sector's biggest operational and margin pressures.

Many grocery chains still use fixed markdown schedules, applying the same pattern of price reductions to similar categories at the same point before expiry. In many stores, staff also continue to audit shelves manually by checking products and dates by hand. That process can leave little time to respond once stock has become difficult to sell at full price.

Retail Insight said its updated machine learning models are intended to move that decision-making further upstream. By combining more granular data with forward-looking signals, the platform seeks to identify likely waste earlier, when retailers still have more scope to change pricing and reduce losses.

Alex Considine Tong, Chief Product Officer at Retail Insight, set out the commercial case for earlier intervention.

"Our Predictive Waste solution helps retailers address their waste and sell-through challenges earlier in the process, when the margin impact is larger," said Alex Considine Tong, Chief Product Officer at Retail Insight.

Smarter markdowns

The feature also feeds store-level sell-through information back into assortment and buying decisions. That means markdown activity is not treated as a one-off response to surplus stock, but as a signal that may show repeated over-ordering or weaker-than-expected demand for particular lines.

Retail Insight argues that this approach could help retailers adjust stock choices in near real time using shelf-level data. In practice, that would give commercial teams another source of information when deciding whether to reduce future orders or alter product ranges in specific stores.

Considine Tong said that feedback loop is part of the product's wider purpose.

"Instead of treating markdowns as an isolated event, sell-through signals can feed back into buying decisions, so teams can see where they have been consistently over-ordering," said Tong.

"Retailers can pull back before the problem repeats and see real shifts in margin and sustainability," added Tong.

The predictive feature builds on the platform's existing dynamic markdown tools, which Retail Insight said are already used to balance waste reduction against margin targets. Within the application, store staff receive prompts on when to prioritise a markdown and what level of reduction to apply under a pre-defined process.

Retail data

Retail Insight said its wider platform now processes 15% of global grocery data across more than 68,000 stores. It added that the platform handled USD $1.4 trillion in grocery revenue last year and delivered USD $795 million in gross profit improvement while helping save 856 million meals from landfill.

The company has spent two decades working with retailers on issues including waste, shelf gaps, stock record errors, and labour pressures. Its customer list includes Kroger, Sprouts Farmers Market, and Marks & Spencer.

The launch reflects a broader push by grocery retailers to use data tools to manage inflation pressures, improve stock discipline, and cut waste at store level, where timing can determine whether food is sold at a reduced price or discarded altogether.

In many food categories, particularly fresh produce, dairy, and prepared foods, the commercial window for intervention is narrow. A markdown made several days earlier can preserve more margin than a deeper discount applied at the last moment, especially where overstocking and weak demand coincide.

Retail Insight's latest update focuses on that narrow window by shifting markdown decisions away from static schedules and toward store-specific signals drawn from trading patterns and inventory data. The company argues that earlier action can change both the financial outcome and the amount of food that ends up unsold.