Demo: Transform to a Modern Tech Stack with Cloud-Native Microservices | Register

How Mobile Premier League Reduced Player Churn with Confluent Cloud

Get started with Confluent Cloud

From Batch to Streaming Data Pipelines

Verfasst von

In online and mobile gaming, the ability to deliver exceptional player experiences is crucial for attracting new players and increasing player satisfaction and retention. This in turn creates a competitive edge that drives revenue generation and long-term success for game developers.

As one of the world’s largest eSports and mobile gaming platforms, Mobile Premier League (MPL) offers 60+ games and hosts hundreds of millions of tournaments every month for over 90 million registered users across Asia, Europe, North America, and Africa. Players can connect with friends or match with other real-life opponents to play and win cash—they expect exciting next-level gameplay that’s rewarding, personalized, and 100% secure with zero performance hiccups.

To support this at scale amid rapid user growth and business expansion with acquisitions, MPL needs to process hundreds of millions of events per day. The lack of real-time data can negatively impact player experiences—when players are mismatched in skill level and experience a losing streak, or if they experience a fraud event, they may permanently leave the platform. Here’s the story of how MPL uses Confluent Cloud to reduce player churn.

Lack of real-time data comes at a high cost

MPL had a batch pipeline system architecture with daily batch ingestion, batch processing, and batch delivery. MPL would receive a daily data dump, then many Spark jobs would run on batch data. This resulted in:

  • Poor matchmaking: Players need to be assigned the right level of play, matched with other similarly skilled players to ensure that they win as much as they lose. Lack of real-time data in matchmaking caused some players to experience losing streaks.

  • Delayed decision-making: Data analysts would run numbers and train models with day-old data, resulting in outdated BI and insights, and the inability to react in real time.

  • Fraud and legal issues: Lack of access to real-time data led to delayed awareness of fraudulent activity, which resulted in negative repercussions.

Together, these contributed to player churn. MPL needed real-time data on a streaming platform, with the ability to build streaming pipelines and do stream processing.

MPL moves to Confluent Cloud

MPL’s goal was to transform their batch-based pipelines into real-time streaming pipelines. They needed to quickly implement a streaming solution and because Jaydeep Punjani, Principal Engineer at MPL, was familiar with Kafka, it was their go-to choice.

However, self-managed Kafka was not feasible. “We only had a handful of people managing our entire infrastructure. We didn’t want to overburden them with also managing the complexities of Kafka. I know how quickly it can get out of control because I've done that in the past,” said Punjani. “Without Confluent managing our Kafka deployment, we’d easily need three or four additional employees to keep things ticking along. But now, we can put those resources into app development.”

MPL seamlessly migrated their entire workload to Confluent Cloud on AWS and Google Cloud.

Leveraging streaming data to reduce player churn

Confluent Cloud provides MPL with a fully managed streaming platform and the ability to build streaming pipelines that stream from any data source (including their gaming app with client and server events, relational databases, marketing SaaS apps) to target systems (e.g., S3) — all in real time. Confluent enables MPL to ingest and process events in real time, scaling to 2TBps+ workloads, over 1B daily marketing events, and 200 million user events/day with 600-700 million events/day at peak during live sporting events.

  • Optimized matchmaking: MPL streams real-time events and player data (e.g., skill levels, scores, win rates) from their gaming app into Confluent and builds stream processing on top of that streaming pipeline to detect when players with a streak of losses are approaching the point of churn. This real-time, enriched view of player interactions gives MPL the ability to react quickly—sending players personalized in-app offers so that they continue to play, while adjusting matchmaking to give them a higher probability of winning the next game, increasing player satisfaction and preventing churn.

  • Onboarding personalization for new player retention: For new players, MPL streams real-time acquisition data (e.g., which campaign they came from, where they’re installing from, social media analytics) from their gaming app and marketing SaaS apps into Confluent Cloud. This data is used to customize journeys in the gaming app with specific tasks for the player. Upon completion, players are rewarded with credits in their gaming wallet. This highly personalized, engaging onboarding ensures new players stay on and don’t feel lost in the large catalog of games.

  • Real-time fraud detection and prevention: MPL streams data from Confluent into its authorization system, allowing the platform to monitor and detect anomalies in real time (e.g., VPN spoofing, multiple logins from different devices and locations, high deposits from new users) and prevent fraudulent transactions, money laundering, and collusion in real time—before they become a problem and cause player attrition.

Confluent Cloud also enables MPL to benefit from increased operational efficiency and cost savings from never having had to hire or reallocate their FTEs to manage Kafka. 

“Confluent streams the data we need to give users complete confidence in our platform, so we can continue to grow and expand our offering while keeping all gamers safe, no matter how they play.”

- Jaydeep Punjani, Principal Engineer, Mobile Premier League

Ready for what’s next

MPL is expanding and acquiring other businesses—these business units are also using Confluent as their central nervous system for data streaming and sharing, increasing the network effect and value from new use cases built on top of existing streaming data pipelines.

As they move forward, MPL is also experiencing deeper collaboration and trust between business and engineering stakeholders, due to the confidence that comes from knowing that real-time data is secure and available when and where they need it. Punjani says, “Our stakeholders have no issue with greenlighting new app developments, as all of our pitches and suggestions are backed up by the data and insights we’ve collected through Confluent.”

To learn more, visit Confluent’s Gaming industry solution page or watch this webinar with additional streaming use cases for gaming. 

Ready to harness the power of real-time data for your business? 

Table of Contents
    • Christine Cignoli started her writing career as an editor at TechTarget and has covered lots of areas of enterprise technology — specifically data infrastructure — since then. She has a master’s degree from Emerson College.

    Get started with Confluent Cloud

    From Batch to Streaming Data Pipelines

    Ist dieser Blog-Beitrag interessant? Jetzt teilen