Company presentation: BlaBlaCar, data for carpooling
Every Saturday we present you a company and its way of using data. After having done an article on data and transportwe come back today to a company that changed part of this sector: BlaBlaCar.
Quick overview of BlaBlaCar
Created in 2006 by 3 Frenchmen, the company BlaBlaCar is a carpooling platform that connects car drivers and passengers. The site was first called covoiturage.fr and has been called BlaBlaCar since 2013. It is one of the first French unicorns, along with Deezer and Veepee.
The platform claims to have around 100 million users in 22 countries, including a large part of Europe (France, Spain, United Kingdom, Germany, Belgium, Poland, etc.) but also in India, Brazil and Mexico.
Since 2011, the platform is paid by taking a commission on the journey.
Starting in 2015, it began to offer new services, starting with offering insurance to its users when they use the platform. In 2017, the platform launched a long-term car rental service. In 2018, the company buys the subsidiary Ouibus owned by the SNCF, which becomes Blablabus.
Here's how Blablacar defines its mission: "We're committed to bringing freedom, fairness and the spirit of sharing to the world of travel."
Use of data at BlaBlaCar
Controlling geographic data
One of BlaBlaCar's great strengths is its mastery of geographic data, which will help it to be relevant to its platform. Indeed, drivers will offer to go from point A to point B, possibly stopping in some specific cities known in advance.
On the other hand, a user will try to go from point C to point D, potentially not far from points A and B of the driver. It is therefore necessary to know the travel times to understand that the driver and the passenger are compatible to travel together. Moreover, with the trip history, we can set up algorithms to know which detours drivers are willing to make depending on the location. For example, if a city is often filled with traffic, drivers will be less likely to make detours of several kilometers because that detour could take more than an hour.
At first, these algorithms were done at the city level, then they were refined to a "points" granularity, i.e. at the user address level.
BlaBlacar has set up an algorithm to propose a price range, depending on the duration and distance of the trip, the presence or absence of tolls, but also the competition with other means of transportation. For example, if there is a fast and cheap train line, BlaBlaCar will lower the price. It will also depend on the day of the week and the time of the trip. A trip is more likely to be taken at a high fare on a Sunday evening than on a Tuesday in the middle of the day - for example, people are closer to work on Sunday evening after a weekend with friends or family.
Users exchange messages among themselves. BlaBlaCar has set up an algorithm to moderate messages in real time. 90% of messages are automatically validated. When the algorithm is not sure enough of the relevance of its filtering, the message is then sent to a manual moderation to validate the moderation or not. The algorithm will then learn from this feedback sent by the human to improve and send less and less messages to manual moderation.
Blablacar is therefore a company that uses data to improve the relevance of its results (geographical algorithms), the user experience (price and geographical optimization) and its productivity (automated moderation)