Johannesburg, 25 Jul 2017
Using algorithms that iteratively learn from data, machine learning allows computers to find hidden insights without being explicitly programmed where to look.
Although many such algorithms have been around for quite some time, current machine learning, which includes the ability to automatically apply complex mathematical calculations to big data - over and over, and faster and faster - is a fairly recent development.
This ability is useful in a number of vertical markets, and can already be seen to be having an impact in the retail environment. The go-to example of machine learning in this space is how major online players like Amazon and Netflix regularly offer recommendations to customers of potential products they might also like. These recommendations are determined via machine learning, which parses through previous choices made by consumers and other products they have looked at on the Web site.
David Cosgrave, Customer Intelligence Lead at SAS, points out that one area where technology is impacting on the retail sector is with customer segmentation. In this space, retailers are constantly trying to break the customer base into smaller, more targeted segments, in order to more effectively market products to them.
"At present, the global trend is towards micro-segmentation, with the ultimate aim being to obtain a segment of one - where each individual customer is treated as a segment on their own. Although locally, we have not yet reached the point where we can create the models to predict at a micro-segment or segment of one level, machine learning offers enormous potential here, and there are a lot of advantages to be gained by implementing this.
For example, he points out that one area where this technology can benefit the retail sector is in the application of machine learning to improve both customer service and the retailers' own efficiencies.
"Machine learning can be used quite effectively in demand planning. All retail stores have to deal with the fluctuations related to different product ranges - for example, in a clothing retailer, they will have to deal with the assortment of colours and sizes that each item comes in. But of course, not all sizes or colours will be applicable to all stores and what is applicable will depend on the consumers in the surrounding community," states Cosgrave.
"Using big data and analytics, coupled with machine learning, the store will be able to determine which colours and sizes are most appropriate for its customer base. In addition, by factoring in additional parameters like the weather in the region and the employment rate in area, the store can ensure that it stocks only the most appropriate items that will appeal to the community."
He explains that this sort of demand planning enables individual stores to ensure they have the right mix of products, for the right customer base, at the right time and in the right store.
Furthermore, by using analytics to take into account things like income demographics, retailers can optimise their ranges to appeal to the broadest market. Such analytics can also help them to predict when the best time is to put items on sale.
While not all of these techniques are radical or even new, it is only the biggest local retailers that are currently moving beyond basic analytics, which help them to understand more about their customers and products, towards true machine learning capabilities.
"Ultimately, machine learning is the gateway to genuine artificial intelligence, so its impact is going to be felt across this entire space. Machine learning is the platform that will enable retailers to reduce their inventory holding, eliminate wasted products on the shelves, significantly boost efficiencies and deliver better customer service. When you look at it that way, the only surprise is that more local retailers are not already focusing on implementing these technologies?"
Share