In the retail and hospitality sectors, data is at the heart of the business. Whether it's AirbnB or Tesco, both operators make massive use of data to generate value.
Whether it's a goods booking or logistics platform, data analysis is critical to growth and performance. Airbnb has based its platform on the exploitation of data, while the retailer Tesco has implemented big data to manage the supply chain of its shops. In both cases, performance is indisputable.
Airbnb uses machine learning to set rental prices
To determine the rental value of a flat, AirbnB uses an algorithm called Aerosolve. This algorithm takes into account a number of variables, including the city, the month of the year, the type of property, transport links, etc. In addition to the classic variables, Aerosolve also analyses images to determine the price. Typically, a welcoming bedroom and a well-decorated living room will be more popular, and therefore fetch a higher price, than a property with any kind of decoration.
While Aerosolve helps rental companies to set their prices, Airbnb has also provided its employees with a platform to help them make requests from non-technical people so that they can make informed decisions. Over the past two years, a third of employees have used this platform, which contains structured and unstructured information, images, data from rental companies, the number of rooms, user feedback, as well as external events such as a festival in a particular town, which will enable the property to be rented at a higher price. Employees draw on more than 1.5 TB of data managed by Hadoop (HDFS) and hosted by Amazon (EC2). AibnB's objective is to work in real time to detect payment anomalies and further personalise the service.
A challenge at a time when the operator's growth is generating ever more data.
Tesco chooses between lemons and limes based on consumer behaviour
With 3,500 shops, 40,000 products and 80 million shoppers a week, the Tesco chain puts consumer buying behaviour at the heart of its analysis in order to be able to provide logistics on 5 continents. While the days on which shoppers are most likely to visit a store are clearly defined, and tend to be at the end of the week, the analysis is more complex than it might seem. The chain knows from experience that each product has a specific marketing pattern. To analyse these sales cycles and fine-tune stock management, Tesco has created a cloud of food products. This enables it to visualise the consumption of each product. For example, lemons are used every day for cooking, whereas limes are sold more at the weekend for cocktails; or broccoli and carrots are sold mainly at the weekend, whereas apples and bananas are sold every day. This visual cloud of data makes it possible to manage weekly sales in a very detailed way, with a predictive approach.
To achieve an ROI of 300% thanks to data analysis, 50 of the 70 employees in the supply chain are data scientists. Most of their day consists of analysing the data stored in a Terradata database and modelling it using Matlab. Thanks to this data work, Tesco is able to manage stocks according to demand and to stock accordingly. As a result, consumer satisfaction has risen by 40%. Another spin-off effect is that, since stock analysis is managed using this model, manual counting by shop is no longer necessary, freeing up staff time to deal with customers in shop.