Growing Engagement on a Feed by 20%

Alter is one of the hottest venture capital backed avatar streaming apps in the world. Alter wishes to keep the advantage Shaped is providing quiet from their competitors who are struggling with personalization – hence we have altered the company name and details for privacy purposes.

Alter was experiencing explosive growth from tens of thousands to hundreds of thousands of monthly active users. While trying to juggle scaling their engineering infrastructure, hiring and building new features they knew that their feed experience left a lot to be desired. In research sessions, users had shared in no uncertain terms that they weren’t being recommended enough relevant or personalized content. As a reference point, users contrasted Alter’s feed to TikTok’s ‘For You’ Page – demonstrating that their expectations were high for personalization.

Fixing the feed had been top of mind for Alter for several months. However, after scoping out the cost of building and maintaining their own in-house ranking infrastructure, it became clear that it was a more significant time and cost investment than originally thought. With the resources available Alter had a decision between building out multiple areas of the app that would differentiate itself from competitors to win the market or almost exclusively focus on improving their feed. Ultimately, they decided to partner with Shaped because they were attracted by the prospect of deploying ‘TikTok like’ real-time personalization in days, the team’s subject matter expertise, and the low maintenance required.

Shaped created Alter’s first model in less than 24 hours to test and evaluate. Within seven days a machine-learning model for personalization was deployed to their feed that leverages Shaped's real-time capabilities to seamlessly adapt and personalize to users' actions in seconds. After deploying Alter meets with the Shaped team monthly to check-in on model performance, feature improvements and additional use-cases they wish to add.

Implementing Shaped's first model resulted in a 20% increase in time spent on Alter’s feed. Users have since shared positive feedback that the feed has been significantly improved.