German ecommerce company Otto uses AI to reduce returns
Otto, the major German ecommerce company, is using artificial intelligence, with an algorithm originally designed for experiments at the CERN laboratory, to improve its online activities. Thanks to the technology, Otto has managed to drastically decline its returns.
Omni-Channel Webinar eTail is the only event where you can learn and network with an incredible group of 80+ senior level retail speakers. Learn from innovative start ups and omnichannel disruptors. With thought-provoking content, interactive formats, and a focus on retail tech, eTail is designed to ensure you achieve omnichannel excellence and drive profit. Download The Agenda
Last week, The Economicist wrote a piece about how Otto is using advanced technologies to surpass the capabilities of its human employees. Using big data and machine learning isn’t new, just look at what retail giant Amazon does. But what Otto does, is something different. Instead of just focusing on recommending the right products to consumers or to personalize the website, Otto is trying to lower the returns of products, which cost the ecommerce company millions of euros every year.
Consumers dislike multiple and slow shipments
An analysis of its data showed that consumers are less likely to return products if they arrived within two days. Also, customers seem to dislike multiple shipments, they rather get everything they’ve ordered at once. But for Otto it’s hard to avoid one of these two annoyances, as it sells products from other brands and doesn’t stock them itself. So it’s either waiting with shipping until all orders are ready or shipping several boxes that arrive at different times.
The solution to these problems is better predicting what customers are going to buy, so these products can be ordered ahead of time. And that’s exactly what Otto did, using a deep-learning algorithm, which was originally designed for particle-physics experiments at CERN in Geneva. It analyzes about 3 billion past transactions and 200 variables, such as past sales, searches on Otto.de and weather information to know what consumers will buy.
90 percent accuracy
The system now predicts with 90 percent accuracy what will be sold within 30 days, so Otto has allowed it to automatically purchase around 200,000 items every month from third-party brands with no human intervention. This system has also led to Otto reducing product returns by over 2 million items a year.
The Economist writes that the project suggests that a major role of AI in retail may be simply to make existing processes work better. “Otto did not fire anyone as a result of its new algorithmic approach: it hired more, instead.”