Wynk Music App by Airtel Raises the User Experience Bar with Amazon EKS

Launched in late 2015 as an innovation unit of Bharti Airtel Limited, one of India’s key telecommunications companies, Wynk Limited runs Wynk Music—a leading music streaming service in India with 72 million monthly active users and more than 14 million tracks in its content catalog.

Wynk’s mission is to be the preferred destination for entertainment in India. Its business goal is to strengthen the monetization of its music app, particularly via paid subscriptions. Wynk’s strategy is to offer personalized premium experiences that help retain existing customers, attract prospective ones, and grow its paid subscriber base.

To grow its music subscription revenue, Wynk sought to offer a high-quality search and discovery experience where users have easy access to their desired content with a single click. Wynk sought assistance from Amazon Web Services (AWS) to launch its new Content Discovery platform, where users are given real-time tailored music recommendations.

After launching its Discovery platform in March 2020, user interaction improved with a 39 percent increase in clickthrough rate to Wynk’s homepage, helping to grow its monthly active user base from 50 million to 72 million. One of Wynk’s top performing products is “My Mixes”. Around 17 percent of daily users listen to “My Mixes”, which accounts for almost 50 percent of their total music consumption.

According to Ankit Srivastava, head of technology at Wynk, AWS helped Wynk move away from manual content curation to an automated and efficient process. Its Discovery platform automatically processes user events, enabling Wynk to customize its music products in real time based on each user’s preferences and in-app interactions.

“Wynk can now offer dynamic elements on each user’s homepage, with five to six different personalized product types to meet their changing content needs. We capture and use up to 60 different user persona attributes such as genre preference, search traits, and usage habits to create very specialized user segments, and match these with various content persona attributes based on existing content and products to offer a highly customized user experience,” explains Srivastava.

To create an agile and horizontally scalable infrastructure to support the new Discovery platform, Wynk moved its monolithic applications to Amazon Elastic Kubernetes Service (Amazon EKS) to easily start, run, and scale Kubernetes applications in the AWS Cloud.

By running AWS EKS, the Wynk DevOps team has around 100 microservices live in production at any time, while reducing development time for new customer services from 2 weeks to 4 days. With a microservice-driven approach using AWS EKS, Wynk’s DevOps teams can run, scale and test tasks independently and in parallel resulting in faster services creation. Utkarsh Sopan, head of engineering, content discovery and personalization at Wynk, states that Wynk enjoys far greater agility than before.

Sopan says, “Our teams can now focus on building impactful customer-related services, because each developer now exercises end-to-end ownership of each microservice. They’ve benefited from using Amazon EKS to easily maintain Kubernetes clusters. We could also fit our existing workloads into half their former capacity and automatically accommodate changing traffic volumes at 99 percentile latency or under 50 milliseconds with the Amazon EKS Cluster Autoscaler.”

Using Amazon EMR to automate and operate its big data environment, Wynk collects seven billion events and processes four terabytes of data on user interactions daily. Hitesh Bhatia, head of DevOps at Wynk, explains how this data is processed using Wynk’s proprietary algorithm.

“Amazon EMR helps us use AI and machine learning models to clean up our data and intelligently power our user personalization efforts with the most effective recommendations. We can exercise greater control over this massive data environment with the flexibility of automated capacity provisioning and cluster tuning. AWS has also been actively involved in this entire journey, providing plenty of support with training, review, and immersion sessions to help us optimize our architecture,” Bhatia adds.

Using deep learning on Amazon SageMaker, a comprehensive service to build and deploy high-quality machine learning models, Wynk is currently exploring a proof-of-concept context-based search and discovery feature—optimizing results based on current context details such as user location and type of device used—to further personalize the app and improve users’ search experience.

Wynk is looking to achieve 300 million active users in the next 3 years, with its services and platform supported by the AWS Cloud. In the near term, it plans to continue migrating remaining workloads to Amazon EKS to further improve user experience and grow its active user base from 70 to 100 million over the next 9 months.

“In the year ahead, we’re looking to further enhance our user personalization with AI and machine learning capabilities for richer personas and increase our paid user base from 2 percent to 10 percent. Since all the services built on our Discovery platform can serve in any content discovery use case, we also successfully optimized our video-on-demand app, Xstream, and have the potential to expand our range of offerings as a leading entertainment hub in India,” Srivastava concludes.

To learn more, visit Telecommunications on AWS