Image recognition technology: seeing the future of big data

Having powered computer vision for driverless cars and advanced robotics, image recognition tech is now being adopted by major fmcg companies to improve in-store execution. By digitising the physical world and collecting data from millions of retail shelves, companies may soon be able to predict how changes to products will affect buying decisions. 

In today’s highly competitive retail landscape, the opportunity to engage the shopper is rapidly diminishing. Not only are there a growing array of products, brands and promotions for shoppers to choose from, but they’re also spending less time in stores. At the same time, deep discount supermarkets are expanding and there’s been significant growth in retailers’ own-brands.

For fmcg companies, being able to control and optimise retail execution at the point of sale has never been more critical. But traditional audit solutions are manual and time-consuming. Auditing takes approximately 15 minutes per product category, involving physical measurements that are prone to human errors. It’s expensive to gather even basic insights. 

Such limited reporting not only hampers timely and effective response to consumer demand and preferences, it prevents manufacturers from making more intelligent, accurate and profitable business decisions.

Image recognition is a game-changer, enabling suppliers and retailers to understand the marketplace in real-time and see what their customers see at all points of sale. They can now instantly process images captured in-store by audit agents. And image recognition can track more than 50 different measurements. 

We’ve run pilots that automate the process - for example by putting cameras on cooler doors so they can continually analyse the shelves of a drinks cabinet. This allows manufacturers to continually assess stock levels and promotional activity. 

Sales reps are using these real-time insights to identify opportunities for upselling, cross-selling and providing range extensions.

With fmcg companies increasingly looking to expand across emerging markets where there are more smaller stores, fewer regimented store shelves and different competitors, automation of store audits has become a necessity. However it’s already being adopted in advanced economies including the UK, where suppliers increasingly need to ensure compliance of purchased shelf space and in-store promotions. 

Perhaps more fascinating, though, is how data coming from the store shelf could be used to predict buying behaviour. It’s widely accepted that 80% of purchasing decisions are now made in front of the shelf. By combining insights from the shelf with PoS data, technologies will be able to link changes on shelf to changes in purchasing behaviour. 

This kind of predictive analytics could not only improve in-store execution, it could also guide product innovation. In a world in which fmcgs are having to constantly evolve their products and with the global retail marketplace becoming increasingly fragmented, data collected from store shelves could become enormously valuable. 

Neil Gowing is managing director, EMEA at Trax