To Fight Food Waste, Grocers Turn to Analytics

Novel ideas and technology stop supermarket profits from ending up in the trash
A moldy yellow bell pepper on a green table.

The next time you’re at your local grocery store, take a drive around back and peek over the lip of a dumpster. What you see might surprise you: more bread, produce and pre-packaged sushi than you could hope to eat in a month.

It’s a typical sight at supermarkets around the world. According to the United Nations, one third of food produced globally is lost or wasted. Yet more than three billion people can’t afford a healthy diet. In Canada, estimates suggest that almost 60 per cent of all food produced is lost or wasted. Canadian grocery stores are responsible for 1.31 million tonnes of this refuse every year.

All those tossed veggies, carbs and meats have a huge impact on the planet—to the tune of eight to 10 per cent of global greenhouse gas emissions. If food waste were a country, it would be the third largest greenhouse gas emitter in the world, behind only the U.S. and China.  

So that’s the bad news.

The good news is that a growing number of tech companies are coming up with solutions to the food waste problem, particularly at grocery stores. They’re using advanced analytics and artificial intelligence to improve quality control, inventory management and forecasting—most with an eye on not only reducing waste but also increasing profits for retailers. 

Bite-sized barriers

Food waste is not an easy problem for these tech companies to solve. The grocery store industry is mature and risk-averse. Thin margins and legacy hardware like computer-assisted ordering systems (CAOs) help explain the often slow pace of change in supermarkets.

CAOs are the most common technology used for inventory management in food retail. But they generally do a poor job forecasting what stores need. To work properly and reduce waste, CAOs must have accurate data inputs, such as delivery times and product sizes. But often this data is inaccurate or incomplete. That means orders are usually based on guesswork.

Still, it’s difficult for retailers to discard these old tools for new technologies, some of which may be unproven. It’s a risk many feel they cannot afford to make.

There are other challenges too, like the regulatory environment (or lack thereof). For example, most retailers manage product expirations using manufacturers’ “best-before” dates rather than ones that could be more accurately set by government or industry bodies. And when retailers try to give some of that “expired” food to food banks (because it’s actually safe to eat), they discover that they can’t. Many countries ban donating past-date products.

So how are tech companies using advanced analytics and AI to get around these challenges? Several ways, it turns out. 

Demand forecasting

A lot of food gets wasted at grocery stores because of inappropriate quality control, product overstocking and inaccurate forecasting. The existing replenishment and discounting models don’t help either, which leads to big losses due to spoilage and expired goods.

Some tech companies have shown that predictive modelling can help, especially with order optimization. With predictive modelling, information like point-of-sale data, weather, news trends, holidays and demographics helps predict product demand.

Some predictive modelling platforms have high accuracy. Take British-based Ocado, which helps retailers like Sobeys sell groceries online. Ocado’s AI-powered smart platform makes millions of accurate predictions per day that reduce food waste. Its software can alert retailers if a stock is nearing its expiration date. At last count, only one in 2,600 products was wasted when retailers used Ocado’s system.  

Computer vision

Most grocers have high standards for the appearance of fruit and vegetables. Items may be tossed if they’re too big, too small, unevenly shaped or slightly off colour. This creates a self-reinforcing loop where customers expect perfect-looking produce. But if these edible-yet-imperfect foods could be sold at a discount, waste would drop dramatically. The problem is that manual inspections eat up a lot of employee time. And that costs money.

Enter machine-learning tools like Eden, which Walmart uses to screen produce for quality and freshness, replacing those human inspections. With Eden, retailers can identify products that may not meet traditional standards but are still safe to eat. This helps them price and discount those products appropriately. Five years after its introduction, Eden has saved Walmart over US$2 billion.

Smart packaging

Assessing the quality, shelf life and safety of food is complicated, especially for perishables like fruit, vegetables, prepared foods and baked goods. It’s no wonder these are the most wasted food items out there. Developers of “smart” packaging are trying to change that by embedding radio-frequency identification and temperature or ripening indicators within product casings. The data can then be analyzed to provide detailed, real-time information on the condition of each product. 

Some retailers are also exploring dynamic shelf-life labelling to accurately determine best-before dates. One of these tools, called Fresh Index, is made by German tech startup Tsenso. With it, expiry dates aren’t printed but read on an app and updated automatically based on production and storage conditions. Recent tests showed that Fresh Index resulted in 30 per cent less food waste due to spoilage and 12 per cent higher margins through quality-related pricing.

Dynamic pricing

Consumers typically avoid buying products that are close to the best-before date. Retailers try to counteract this tendency by discounting these products. But as with manual inspections, discounting individual items takes a lot of employee time. With dynamic pricing, however, digital displays lower prices automatically as best-before dates near. When combined with predictive modelling tools, product forecasting can drastically improve. 

Israeli startup Wasteless is a good example of how this can work in practice. It uses a machine-learning technique called “reinforcement learning” to power its dynamic pricing engine at grocery stores. The algorithm uses data, such as the arrival date of the next product shipment and the number of units remaining in stock, to avoid underpricing or overpricing. It has been shown to reduce food waste and boost revenue for several retailers.

Scan-based trading and consignment

This is a new business model emerging in the grocery industry whereby the vendor assumes responsibility for the inventory and the retailer only pays for products that sell. When scan-based trading and consignment is paired with predictive demand forecasting and automation, retailers can expect both a reduction in food waste and a boost in profits.

Shelf Engine is leading the charge in this space. The company’s platform connects daily inventory data and other real-time information from stores with machine learning and analytics to create highly accurate orders. The beauty of the platform is that no new hardware or system changes are required, and any products that aren’t sold are repurchased and donated or repurposed. One national grocery chain in the U.S. that uses Shelf Engine saw profit margins increase to 25.6 per cent from 15.6 per cent.  

Feeding forward

The global food waste problem can no longer be ignored, and grocery retailers are realizing the essential role they play to solve it. Walmart Canada, for instance, has committed to zero food waste by 2025 and Costco has a goal to divert 80 per cent of its waste from landfills.

To hit these goals, retailers should consider the types of analytics-driven solutions described here. Not only can they help drastically reduce food waste but some can also boost profits.

Recent research co-authored by Anton Ovchinnikov, Distinguished Professor of Management Analytics at Smith School of Business, shows that statistical model-based inventory management systems do a far better job with inventory decisions than human-based decision-making. The research also shows that analytical-based approaches to stocking shelves can result in a seven to 10 per cent increase in profits.

Of course, there is a moral imperative here too. The UN calculates that if food waste was reduced by just 25 per cent, there could be enough food to feed all food-insecure people globally. Not only will reducing food waste directly impact retailers’ bottom lines, it will also have implications for the health of our world.

 

This article was researched and written by a team of Global Master of Management Analytics students at Smith School of Business: Amol Gupta, Shirley Hu, Zongying Hu, Olugbenga Ilori, Patrick Linehan, Jessica Niles, Hari Saripalli and Jiawei Zhang.

Subscribe to the Insight Newsletter

Keep up with the latest in Smith thought leadership, faculty research, and more.

Subscribe

Smith School of Business

Goodes Hall, Queen's University
Kingston, Ontario
Canada K7L 3N6

Follow us on:

Queen's logo