AI-enabled freezers help boost ice cream sales

With nearly 3 million of its ice cream freezer cabinets placed in shops, airports and supermarkets around the world, Unilever is now using AI technology to help meet consumer demand. The AI tech has been installed in around 100,000 of the ice cream freezer cabinets globally, and the company has plans for this to be increased to 350,000 by the end of 2025.

The software in each freezer unit is carefully positioned to only take photos of the stock inside the freezer. The images are taken periodically and automatically sent to the cloud, where they’re analysed by AI that’s trained to only identify Unilever’s ice cream brands, analyse their depth in the cabinet and use an algorithm to generate an order for the distributor.

“It’s like two eyes are looking at our stock every day,” said Sarosh Hussain, Ice Cream’s Head of Digital Selling Systems. It also means whenever a retail outlet asks for stock, the distributor knows from the data exactly what their freezer needs.

So far, the uptick in retail orders and sales thanks to data from the AI-enabled freezers has been significant.

“There is no one size fits all. It varies from country to country,” Sarosh said. “But we have seen incremental growth that ranges from 8% in Turkey to 12% in the US and up to 30% in Denmark.”

In Hungary, where almost 10% of its freezer fleet has been AI-enabled this summer, work has also been done to integrate the freezer’s system with those used by distributors.

As well as seeing the retailer’s regular ice cream order, distributors receive an order recommendation based on images and AI analysis of the stock levels in the cabinet.

The response from both distributors and retailers to these ‘recommendations’ has been positive. So far, close to 5% of orders have been generated from the AI system’s input this ice cream season.

Every ice cream order is fulfilled by refrigerated vans. Sarosh noted, “If you have a more accurate understanding of which customer wants what, distributors can plan their tertiary route and distribute to the outlets efficiently and more cost-effectively.”

Big data from its freezer fleet will also give the company clearer insights into the performance of new product launches as well as marketing campaigns.

“By analysing this data, we should be able to understand if our ice cream brand’s above-the-line campaigns have had a positive correlation on freezer sales and stock,” Sarosh said.

“We’ll be able to compare freezer sales from previous seasons to determine if new products are adding value or cannibalising our core, and have the insights to reassess promotion planning and reforecast accordingly.”

Image credit: iStock.com/Aleksandar Pirgic