Is it time to replace your digital merchandiser with a machine?

Dimitri Sotiropoulos

7th Jul 2016

Retailers’ endemic cultural resistance to automation and machine learning is the biggest challenge to its integration, not the tech, according to Jon Bovard, e-commerce director at UK online beauty retailer, Cultbeauty.

Speaking at the recent Inside Retail Academy event, Bovard said that the answer to what drives online sales outcomes today is an entirely different proposition to five or 10 years ago.

An MIT study of 168 large companies with $500 million revenue or more, showed three quarters of them were targeting higher sales growth from machine learning. “38 per cent credited improvements from sales metrics… machine learning is here to stay and changing how we interact with customers and they way that we market, sell to customer’s and operate logistics,” said Bovard.

Typically, an omnichannel business that implements personalisation and automation across omnichannel initiatives will increase sales revenue by 5 per cent, Bovard said. Automation of merchandising represents a ripe area for capital gains, with the average visitor of any website looking at less than one per cent of products.

“Many of us invest hundreds of millions of dollars in tens of thousands of SKUs, and yet visitors are making a purchase decision — yes or no — based on less than one per cent of products,” he said. “That scares people, because we have all these amazing products, you come to our site and look at three or four of them and decide whether or not to buy.

“That right there highlights exactly why machine learning, targeting and recommendations through personalisation have a place in retail, because we only have a very limited amount of time to connect someone with a product they are more likely to purchase.” 

This is a contentious issue for merchandisers in retail, said Bovard, who don’t fancy the prospect of relinquishing control to, essentially, a computer.

Citing an example of a retailer who was employing someone to merchandise their digital properties, Bovard said the employee was threatened by handing over the reins to the machine. The employee moved from “doer, to analyser.”

“All of a sudden his job became more about understanding metrics, understanding what customers buy and [how they] engage,” Bovard said.