DenmarkMedia and entertainment2025 – ongoingPodimo
Personalized recommendations for podcast and audiobook discovery
Podimo's personalized playlists beat hand-curated ones by 15% and cut churn by more than 3%
Problem
With more than 2 million podcasts on the market, listeners struggled to find shows to start with. Recommendations on content platforms often lean toward popular titles, so new or niche podcasts get overlooked.
Approach
Podimo's AI team trained several recommender models on Vertex AI and ran them on Google Kubernetes Engine. Different strategies serve different places in the app, from simple content-based ones to a deep learning model. The service computes personal recommendations for every user daily and also works as an API that re-ranks any collection in the app for each user.
Results
Takeaway
Podimo compared personalized collections directly with one-size-fits-all curated ones on conversion and engagement. The company said engagement-only training objectives did not give the best long-term business results.
Categories
- Industry
- Media and entertainment
- Business function
- Marketing and sales
- Innovation type
- Product and service innovation
- AI technology
- Recommendation and personalization
- Adoption
- Built in-house
- Evidence
- Vendor case study
Sources
Source summaries
Google Cloud's customer story describes how the Danish audio subscription service Podimo built its own recommender models on Vertex AI. It reports that personalized collections performed 15% better than curated ones, churn fell by more than 3% and recommendation diversity rose by up to 30%. The page carries no date; the Internet Archive first saved it on August 28, 2025.
Full details are in the original sources.
