Ocado, the world’s largest online grocery retailer, employs machine learning (ML) to fight shopper fraud. Its warehouses run on artificial intelligence (AI) and ML-algorithms for their routing systems. In addition, its AI-powered Ocado Smart Platform forms the backbone of other online retail businesses worldwide. These three business branches have one common denominator: they leverage cutting-edge technology to ultimately create the optimal customer experience. Please read along to learn how Ocado’s branches fulfill this goal.
A classical online retail problem: orders are paid and delivered, after which a refund is requested. This generally occurs when the credit card used was a stolen one. With a fraud rate of 0.1% of orders, fraud is quite rare. However, it poses a significant threat to Ocado. If left undetected, fraudulent information could spill into other systems and business divisions, impacting customer service. A highly precise fraud detection model is critical. A 99.9% accurate ML-model would still miss multiple fraud occurrences. Therefore, Ocado developed a top-notch ML-based algorithm to reliably detect fraud among its shoppers.
How? By implementing a deep neural network on TensorFlow, Google Cloud’s open-source ML library. Ocado analyzes various data points to train their model as best they can. "From the data we had collected from past orders, including cases of fraud, we created a list of features which included the number of past deliveries, the cost of baskets, and other information," states Holly Godwin, Ocado's PR Executive. The results speak for themselves: Ocado’s fraud detection precision has improved by 15x. For more details, please read this article or watch this video.
Ocado’s warehouses are full of robots picking goods to include in an average of 260,000 orders a week. The core of the magic: vision systems and artificial intelligence. These enable the robots to process orders as quickly, carefully, and efficiently as possible. For an impression of the complexity and extent of the warehouses, please see the video below.
Ocado tackles various warehouse challenges with technology. For example, robots needed to learn where to find the barcode of products in order to identify and pick them. In the end, the entire barcode location step was bypassed by using image recognition technology. This expedited the process and improved its accuracy. Here, too, TensorFlow was used to build the ML-models.
Now social distancing has become the norm, some of Ocado’s automation products and techniques have gained interest from other parties. This brings us to the Ocado Smart Platform. This complete end-to-end system allows grocery retailers to expand their market share in the fastest-growing retail channel in the world. Ocado Smart Platform can create an intuitive interface for online shopping, build an efficient order picking system with automated facilities and in-store picking systems, and enable fast and convenient delivery through proprietary routing systems and algorithms. With a real-time understanding of data generated across the order process, the Ocado Smart Platform lifts the customer experience to the next level. Customers benefit from an easy-to-use online store, the supply chain is as efficient as possible, and the technology truly works for the employees.
With innovations in cloud computing, AI, big data analytics, and ML, Google Cloud enables Ocado and Ocado Smart Platform to push every aspect of online grocery retail further. To benefit from technology as well, please visit the Google Cloud website and Ocado’s website for more information.