Behind the Scenes of Wrapped: Serving Personalised Data Stories at Scale
Fred Wang is a Senior Backend Engineer on the Mambas team at Spotify's New York office. His work on 2021 Wrapped centred on delivering data story content for iOS and Android, translating raw backend data into the personalised, localised experiences millions of users saw at launch.
From Raw Data to Personalised Storytelling
The Mambas team owns all animation, presentation, and personalised data within the Wrapped interactive mobile app. Wang's responsibility is to take raw data from the backend, enrich it, and then translate and format it for different regions before it reaches frontend mobile engineers. This includes determining which music plays during data stories, selecting images, and localising text — essentially everything served from the backend to ensure iOS and Android can render the stories correctly.
The Hardest Part: Simulating Peak Load
The most challenging aspect was load-testing. In the first few hours after launch, Wrapped receives a high volume of unique visitors, each with their own data story served to different mobile devices. This means the team cannot simply deploy and hope for the best; they must simulate that load to verify the system can withstand requests at peak usage.
During load-test sessions, the team processes high volumes of requests and attempts to uncover as many issues as possible across their dependencies. It is a substantial effort, but one that reveals every weak link in the chain.
Launch Day and Team Highlights
In the first two hours following launch, traffic was unprecedented and global — yet the system scaled without any major issues. While nerve-wracking, it was immensely satisfying to see the preparation pay off.
Wang also emphasised the collaboration across engineering and localisation teams to make data stories work smoothly in every region, language, and format. A standout moment was when the data pipelines first went live and the team enabled live data for their endpoints, allowing everyone to see their own Wrapped stories for the first time. The excitement was tangible; for hours afterward, the team played with the feature and compared notes.
