Yotpo: Leading eCommerce retention marketing platform enhanced with AI
Yotpo is an eCommerce marketing platform that helps brands drive growth by creating engaging experiences to build lasting customer relationships. Yotpo's integrated solutions for reviews, visual User Generated Content (UGC), rewards, and referrals empower businesses to win over new audiences using their customers' voice.
Challenges
Intensive Engineering Support: Crafting recommendation engine models demanded significant engineering backup.
Infrastructure Scalability: The infrastructure needed to be robust enough to support billions of users concurrently.
Platform Requirements: Yotpo sought a real-time ML model serving platform that was reliable, user-friendly, scalable, and forward-looking.
Flexibility & Autonomy: It was pivotal that Yotpo remained unshackled by specific model structures. They also required a centralized mechanism to seamlessly train, deploy, and access models on a massive scale.
Operational Overhead: Building and maintaining such infrastructure demanded extensive expertise, time, and effort—areas not aligned with Yotpo's primary business objectives.
Implementation
Solutions
Seamless Integration: Integrating with Qwak was a straightforward affair. A simple wrap-around of the Python model and a few lines of code empowered Yotpo with a fully deployed system. Notably, Qwak's platform promises flexibility, facilitating quick implementation of novel features.
Visibility with Qwak Console: A fully integrated UI ensures Yotpo can comprehensively monitor all deployed models. Moreover, this console integrates seamlessly with Opsgenie, Yotpo's primary alerting system.
Scalable Model Deployment: Qwak's solution caters to both real-time and batch model deployment. The inclusion of canary deployment ensures Yotpo can effortlessly handle millions of real-time predictions.
Autonomous Feature Management with Qwak Feature Store: This feature offers a streamlined API for feature extraction and retrieval, granting the data science team autonomy. They can now tweak model inputs without backend modifications. Plus, sharing feature sets among various models and conducting large-scale model training become feasible.
Analytics & Insights through Model Monitoring: It becomes possible to review prediction data and craft intuitive dashboards overlaying real-time data. This provides invaluable model insights for Yotpo's business decision-makers.