Betclic: Putting Data Science at the Center of Online Gambling

The global online gambling market is anticipated to be valued at more than $94 billion in 2024, and increasing use of more advanced data science, machine learning, and AI techniques to improve the online gambling experience will most certainly play a role in this market growth. 

In fact, one of the industry’s biggest challenges could be what to do with all of the masses of data at their disposal and whether they can turn all of that data into actual business value. Unlike traditional sales use cases, data science in online gambling means not only personalization and anticipation of individual behavior, but also taking into account the uncertainty surrounding any action the user takes. We spoke with Bertrand Warot, Data Science Manager at French online gambling company Betclic, about how they are rising to these challenges.

About Betclic

• Founded in 2005 and currently the largest betting company in France.
• Specialized in sports betting, horse race betting, online casino, and online poker.

Some examples of use cases the data team at Betclic has worked on include:

• Anticipating new customers’ future value 
• Ad-hoc analysis on data from A/B tests
• Money laundering detection
• Linking related fraudulent accounts

One of their biggest use cases is helping business teams better understand the customer overall through advanced customer segmentation (including looking at whether users are active or not, new accounts or older accounts, and how long users stay idle before returning) and — for marketing purposes — trying to predict the customer’s next action. 

Development of a Data Department

Yet tackling a wide range of use cases and attempting to put data science at the center of all operations didn’t come naturally — it required some fundamental organizational change. Betclic’s CEO created the data department in 2017 in an effort to do just that — make data the core of all Betclic projects. 

Since then, the team has grown from just three people in the R&D department to a proper data department of more than 30 people who are responsible not just for one-off projects, but the overall management and incorporation of data into processes. The department structured around three areas: data engineering, analytics, and data science, all of which sit under a Chief Data Officer (CDO).

“In the context of rapid expansion of the data department, Dataiku supported us in upskilling analysts while also enabling us to ensure the same level of productivity with the already-established team.”

– Bertrand Warot, Data Science Manager at Betclic

Processes and Keys to Success

Along with the organizational change required to upskill people and staff a data department, Betclic was able to better tackle some of its use cases by making some process changes as well. 

Before: In the early days of data science efforts, access to data was controlled through written request to the database administrators, after which they would deliver datasets as CSVs (which may or may not correspond to expectations or the data needed for the project).  Each team member worked independently with his or her own processes, organization, and versioning, and files were stored on local machines.

…vs. After: To scale their team, the data department needed a way to eliminate maintenance and communication issues caused by their previous setup. Betclic turned to Dataiku to help centralize all data efforts plus offer a solution to sharing of projects, versioning, archiving, and automation. As the team continues to grow and junior analysts come with new skills, Dataiku enables Betclic to centralize all of that work and train staff to handle more (and more complex) parts of the data-to-insights process.