About
A leading podcast platform partnered with Dataforest to replace manual recommendations with an AI-powered personalization engine, resulting in 7× higher user engagement and significantly increasing revenue.
Challenge, approach, and impact
Personalized Recommendations Model Architecture
Created a flexible model architecture allowing any item (podcast episode, comment, or metadata) to serve as a feature or recommendation target, enhancing relevance and engagement across the user journey.
Scalable ETL Pipeline Development
Developed a scalable ETL pipeline to automate data extraction, cleansing, and transformation, ensuring reliable, high-quality data feeds into the recommendation engine for improved personalization accuracy and consistent real-time insights.
Hybrid Approach for New User Engagement
Engineered a hybrid approach combining contextual metadata with sparse user data and similar user behaviors to provide relevant recommendations even for new users with minimal history, boosting engagement and accelerating adoption.
Scalable System Architecture Design
Suggested an optimal architectural approach and built a modular, automated system designed for seamless scalability and flexibility. Integrated multiple recommendation modules targeting specific user traits or content types, and finalized results using a learning-based ranking model for high-quality recommendations as the platform grows.
How we built
Testimonials
John Doe, AI Specialist @ Dataforest
Dataforest
“John Doe, an AI Specialist at Dataforest, shares the success story of how their AI-powered personalization engine led to a 7× increase in user engagement for a leading podcast platform.“
Team structure
Client team
Client Representative Representative
Project stakeholder
Project stakeholder
The client stakeholders at THAMANYAH were working closely with the team at Dataforest
Agency team
3 x Data Engineer
Production
2 x Machine Learning Engineer
Production
1 x Business Analytics
Production
2 x Product Manager
Production
