CASE STUDY

Streaming Platform: ML-Driven Personalization on Google Vertex AI

A streaming media platform built ML-driven personalization on Google Vertex AI. Watch time per session up 22%; subscriber churn down 11%.

May 9, 2026 · 2 min read · By admin

At a glance

Industry: Media
Topics: AI / ML · GCP · Predictive Analytics
Published: May 2026
Headline outcome: +22% Watch time per session: +22%

Measured outcomes

What this programme actually delivered.

+22%

Watch time per session: +22%

11%

Subscriber churn: down 11%

The challenge

A streaming media platform’s recommendation engine ran on legacy heuristics. Recommendation quality lagged the personalization-leaders in the streaming category.

What we built

Vertex AI-based recommendation models. Multi-objective optimization (engagement + content-discovery + retention). A/B-testing platform integrated. Real-time scoring at sub-100ms latency.

Outcomes

Watch time per session: +22%
Subscriber churn: down 11%
Content-discovery breadth: significantly improved (engagement spread across catalog)
A/B-test cycle: weekly model improvement cadence