Delighting millions of social gamers worldwide

Improving user retention with real-time, actionable data modeling

In mobile gaming, winning is everything. SciPlay, creator of Jackpot Party, one of the top social casino games in the world, offers gameplay and the opportunity to beat the house every day to more than 6 million monthly users with games that are highly engaging, unquestionably secure and always dynamic. In response to the pandemic and recession-driven disruption, SciPlay has built its core business initiatives around adaptability, accuracy and a commitment to user safety. Rising cybercrime, coupled with an accelerated focus on profitability and customer retention, has fueled the need to be more data driven and thoughtful about how processes, activities and investments impact ROI.

However, SciPlay’s cloud data warehouse was complex to manage and costly to scale as the company and its data grew exponentially. A growing portfolio of games, millions of daily users and sessions, and internal data from multiple departments led to countless delays, thanks to the complicated and time-consuming process of manually running ETL and building pipelines to support analytics and AI initiatives, including strategic decision-making, customer segmentation, churn prevention and revenue forecasting.

Since SciPlay’s existing data warehouse required so much DevOps support, the overall infrastructure became a patchwork of tools and services left over from point solutions and acquisitions. Disconnected data silos blocked access to key information and made sharing across teams a struggle. Collaboration disintegrated, and without the ability to extract insights from ballooning data inputs, efficiency and performance took a hit. “Instead of reacting in minutes, it would take us hours to make a critical decision,” said Shmuel Ben-Meleh, VP of Global Analytics & Data Science at SciPlay. “That was not acceptable for us as a real-time, always-on business.” This is where Databricks came in.

Democratizing access to data, analytics and AI to facilitate all facets of business

SciPlay needed to modernize its data infrastructure with a focus on facilitating efficient workflows, building streaming ETL pipelines that were reliable and fast, developing machine learning models to improve gaming experiences, and providing actionable insights to teams to drive the business forward. With the implementation of Databricks Data Intelligence Platform, the team successfully unified all of their data, analytics and AI within a single platform, simplifying operational processes and enabling broader access to data and insights across the organization.

“Data is one of our most precious assets to inform our decision-making,” said Ben-Meleh. “Coupled with our collective experience and knowledge, it’s a very powerful combination. But we couldn’t leverage all of what we had with a siloed warehouse and disconnected tools.”

With data and analytics fully democratized, SciPlay can now support various business teams from finance to marketing with reports and dashboards built in both Tableau and Databricks SQL. “With actionable information at our fingertips, nontechnical users can now access data and generate insights easily and independently,” explained Ben-Meleh. The data science team is also benefiting from the ease of data access, building models within MLflow to target cohorts of players, and serve personalized experiences to increase engagement and retention.

Finally, SciPlay leverages Databricks Data Intelligence Platform to round out the value of two separate internal centers of excellence. The first is dedicated to technical best practices and providing proper infrastructure for data teams; the second is to democratize data and analytics for business teams and non-technical users. The returns have been swift and significant, creating dramatic cost savings around decision-making, and paving the way to develop tools focused on improved operational efficiency.

“The texture of our data is much better with lakehouse architecture,” said Ben-Meleh. “Conducting exploratory analysis between multiple data sources used to take several days — even just to gather the data prior to analysis would’ve taken that long. Now we have one place to go for a complete picture of the data, and can generate insights much, much faster.”

Accelerating time-to-launch with smarter, faster, more accurate decisions

As a gaming provider, acquiring new users is extremely expensive, and keeping existing users playing on the platform is crucial to maintaining growth. Now, several of SciPlay’s machine learning models produce remarkable results with Databricks Data Intelligence Platform as its foundation, particularly its new churn prevention model, which has allowed it to identify customers at risk of churning so that it can determine strategies to improve satisfaction and retention. Since adding lakehouse to its tech stack, SciPlay has seen a 5% uplift in user retention across multiple games and can now launch new games an average of 75% faster than before.

Internally, non-data teams benefit from vastly improved time-to-insight, which Databricks helped reduce from 4 hours to just 4 minutes — a 60x improvement. A more efficient lakehouse architecture has deeply improved operational efficiency, drastically lowering costs and helping SciPlay to deprecate its legacy database footprint, further reducing capital expenditure. Finally, engineers now have a far more robust infrastructure in which to stream data — outputs are more secure and reliable, analysis is deeper and more complete, data has become fully actionable, and workflows across platforms have accelerated as a result.

“Databricks Data Intelligence Platform has been a plug-and-play solution that has helped us level up our ability to use data to develop new games that will delight customers for years to come,” concludes Ben-Meleh.