Search is fundamental feature of mobile.de platform and we as Data Team work hard to improve user experience by bringing personalization and relevancy into search flow. Hardest challenge is to cater to user interests while still displaying sponsored ads and obeying dealer interests. To solve it our team created and combined a number of data-driven products such as user profiling based on browsing history, car price ratings based on historical data from mobile.de inventory and ML model predicting click probability based on user tracking data. I will describe each of the products, highlighting the architecture and show how each of them fits into a big picture and integrated with existing search flow. Also I will share main learnings and outcomes.

This talk is presented by ebay tech.