Interview: Otto combining machine learning with people power

Predictive analytics technology provider, Blue Yonder, has been running its European Data Science Academy for two years – and it is now looking to offer more dedicated courses to the retail industry on how best to manage big data to help its businesses run more efficiently.

With courses aimed at top-level management and less senior job roles alike, as well as those individuals who simply want to become data scientists, there is an expectation among Blue Yonder's management team that a growing number of people and organisations will seek this education route in the near future.

Professor Michael Feindt, founder of the tech company, told this publication that there are “many companies” which do not yet understand that knowledge of data science “will be of extreme importance for survival”, and even if they have recognised its importance, there has so far been a reluctance to address the issue.

Blue Yonder does acknowledge, however, that compared to other sectors retail has been quite advanced in its use of big data science, thanks in part to the industry being defined by small margins and ever-growing competition for sales.

One retailer – and a Blue Yonder partner – to have placed data analytics at the heart of its operations is Germany-based mail order business, Otto Group. The company has been working with Blue Yonder since 2006, taking advantage of a number of tools offered by the vendor.

Michael Sinn, Otto's vice president of category support, who has responsibility for planning, forecasting and process optimisation at the retailer, told Essential Retail that it is now vital for people to work alongside machines, to achieve successful results in retailing.

“The business is getting more complex every day and we have a lot of influences coming from big data, social media and the fact competitors are only one click away,” he explained.

“No-on can really make a correct decision, you have to have machine support.”

He added: “We try to let the systems make more decisions – it's not fully automated, we also use human beings – our employees – to check things [before deploying a strategy based on the machine learning]. In the long run we need more skills based on knowing about IT systems and processes. We are in a change process.”

Sinn said that the Blue Yonder technology used by Otto, which includes specific solutions for forecasting and dynamic pricing, has helped contribute significantly to bottom-line gains at the organisation.

The predictive analytics algorithm implemented in the business allows Otto, which sees around 85% of its sales generated online, to constantly update prices based on data and additional factors such as moves made competitors, market trends and weather conditions. A price optimisation solution tests and measures the relationship between changes in price and demand.

Since working alongside Blue Yonder, Otto has reported double-digit growth in its sales and margins, with strong increases in its womenswear lines in particular.

“It helps us to increase the quality of our forecast dramatically,” Sinn noted.

“For Otto, this is very important because we have two million different items online across over 4,000 brands and nobody can really estimate how many shorts in colour blue and size 41 you will sell so we need some systems.”

Blue Yonder is not the only company providing analytical technology for retailers, of course, with players such as 4R Systems now extending their global reach and showcasing how a scientific approach to profit targets can generate positive results. Otto's ten years of work with Blue Yonder, though, is indicative of the importance it attaches to the partnership. The data science company was formed on the basis of the work conducted by Professor Feindt at CERN, the European organisation for nuclear research, where he developed the NeuroBayes algorithm for the analysis of experiments in elementary particle physics during the hunt to discover new particles.

Part of Sinn's role at Otto also covers corporate social responsibility, and the reported improvement in forecasting at the retailer has apparently led to fewer products travelling by air to Otto's distribution hubs around the world.

“We can now use ships or trains to import in other markets, and we have reduced the number of flights dramatically,” he remarked.

There is certainly a lot of talk about data analytics and the importance of science in today's retailing world in retail, but as Professor Feindt suggested, there is still limited movement by the wider industry to address this particular skills gap.

Grocery e-tailer Ocado is one of the exceptions, with chief technology officer Paul Clarke telling Essential Retail this week that he is actively recruiting staff that can get to grips with the robotic technology, machine learning and new system opportunities that exist within his business. He also suggested the very fact that this kind of technology is being used by Ocado makes the organisation an attractive place to work for today's graduates and engineers.

Companies like Ocado and Otto recognise how people power and the most advanced machine-learning capability can sit side by side to bring about retail success. And it is expected that many others will soon start to identify this, too.

Click below for more information:

Blue Yonder

4R Systems