H&M on using AI to become a data driven company

H&M is one of the most recognisable global retailers, with 5,000 stores in 74 markets and flagship shops on the world’s busiest high streets. The company has come a long way since it was founded in 1947 in Sweden.

But more recently it has made a number of "significant" investments in order to meet “the rapid shift in fashion retail,” according to its last annual report. That has included investment in “digitalisation, a more efficient supply chain… and in tech infrastructure, advanced analytics and AI.”

In fact funding for AI has become a key component of its transformation plans over the last few years, according to Errol Koolmeister, head of AI at H&M’s Advanced Analytics and AI unit. “H&M’s machine learning journey started quite late. It wasn't until 2016 that we started proof of concept and tested our advanced analytics appetite and applicability,” he told delegates at the Virtual AI Summit, during London Tech Week. But the function quickly grew across the business.

By 2018, the AI unit became a separate entity within the organisation - “the first new function in 10 years” - with more than 120 people. Since then it has rolled out a number of pilot projects, which have become core components of the business. Now the focus is to reshape H&M to be a “fully data driven” organisation.

Cross-business functionality 

“We are covering the entire value chain, we are doing everything from design buying support, with assortment quantification, fashion forecasting into production, where teams… are looking at how we can support the productisation of the clothes we are ordering. Logistics: where will the clothes go?… Then also in the sales cycle: making sure we optimise the reduction and discount set-up within H&M across the world. And then in marketing: personalisation supported by the H&M advanced analytics landscape.”

It’s a lot over just a few years: “We've done so many things in such a short time at H&M.” But in order to make it work at scale, reuse of the technology will be key. “And if we do that then we don't need 1,000 people… We need more specialists to create the tools and technologies we need to be able to take the next step.”

There are also a number of things to be mindful of when using AI, such as being aware of biased data and keeping track of that as well as guaranteeing the quality of the products the unit produces. 

And it remains important that the AI function doesn’t become a separate silo within the business. “How do we democratise AI? We need to bring tech closer to the business side… [to] incorporate it into the solutions they want to build.”

Koolmeister notes that while the company’s AI journey started relatively recently, the idea of it being a “data driven” company was already there from the start when the company’s founder said: “We will be where the money and the bags are.” It's just the transformation of retail means the money and bags are no longer just to be found on Fifth Avenue in New York or Oxford Street in London.