Product descriptions:
The mobile gaming market is more saturated than ever, making it increasingly difficult to launch new successful titles. Supercell, the creator of hits like Clash of Clans and Brawl Stars, aims to build games that are played for years and remembered forever, drawing inspiration from cross-generational franchises. To achieve this, they need to blend the art of game design with the science of hard facts. Supercell turned to the Databricks Data Intelligence Platform to democratize data access, simplify data processing and enable everyone from game designers to product managers to make hypothesis-driven decisions that keep nearly 300 million monthly players engaged.
Enhancing velocity in an ever-evolving gaming environment
Supercell’s mission is to create games that last, which requires keeping live games fresh and interesting based on player feedback and behavior. However, their previous data architecture created bottlenecks. The team was self-hosting Apache Spark, which became a significant burden to manage. Upgrading to new versions was difficult, leaving the company stuck on older versions and unable to utilize new features that data practitioners requested.
The primary challenge wasn't data silos – data was already collected into a unified location – but rather velocity and user experience. Time-to-insight was slower than desired, and the performance of data pipelines needed improvement. Additionally, the team relied on a self-hosted solution for notebooks that did not provide the seamless user experience required to truly democratize data across the company.
"We discovered that velocity, specifically time-to-insight, wasn't where we wanted it to be," says Boris Nechaev, Head of Data Platform and Engineering at Supercell. "We were spending too much time managing infrastructure instead of supporting our games. When we started looking for new solutions we were looking for someone who could manage Spark, someone who could help us run it properly so we could focus on scaling. Who better than the creators of Spark to do it?"
Democratizing data for game developers and analysts alike
Supercell migrated to the Databricks Data Intelligence Platform to solve these infrastructure headaches. The impact was immediate: Spark and a superior notebook experience came "out of the box," resolving their core velocity issues within months. This shift allowed Supercell to focus on their broader mission: making it easy for everyone in the company to work with data.
Today, every internal employee at Supercell has a Databricks account by default. This open access fosters a culture of curiosity and hypothesis-driven decision-making. Nechaev shares a prime example of this democratization in action: "I recently saw an artist, a persona who doesn’t often engage with data, analyzing player data via a link to a Databricks notebook shared on Slack. They could immediately see the thought process and the journey behind the insights."
By moving to Databricks, Supercell has been able to marry the "art" of game creation with the "science" of data. Game designers and operators can now easily formulate hypotheses such as how to balance a character or design a feature, validate them with data and deploy changes that make the game more fun for players.
Future-proofing governance with Unity Catalog
With over five petabytes of data and a culture of openness, governance is critical. Supercell utilized Unity Catalog to implement a robust, future-proof access control system. While the company encourages open access, they strictly protect Personally Identifiable Information (PII) to comply with GDPR and other regulations. Unity Catalog enabled them to manage these restrictions seamlessly while keeping the rest of their data accessible.
The migration to Unity Catalog also served as an opportunity for "spring cleaning," allowing the team to deprecate unused datasets and standardize on Delta Lake. Moving from Parquet to Delta Lake forced data consistency, trading the short-term effort of migration for long-term reliability and performance.
"Migrating to Unity Catalog was like deleting legacy and stepping into a new universe," concludes Nechaev. "It allowed us to build a proper access control system that we didn't have before. Now, we have a foundation that is secure, compliant, and ready for whatever new features Databricks releases next."