"The recommendation is not a feature. It is the product."— S+3 Agile Field Record
The Challenge
Prime Video's global expansion required more than infrastructure scale — it required intelligence. A catalog available in 180+ countries is worthless if customers cannot discover what to watch. The recommendation surface was the product, and it had a measurable problem: content-quality signals were absent, meaning the algorithm was flying blind on what actually constituted a good title for a given customer.
The ask was to build and launch the Prime Video platform across 180+ countries while simultaneously solving the content-intelligence gap. Both problems had to be addressed in parallel — infrastructure scale and algorithmic quality were not sequential, they were co-dependent.
The Intervention
Title Quality Scoring algorithm
Himanshu established the first Title Quality Scoring (TQS) algorithm and platform at Amazon Prime Video — a system to measure, rank, and surface high-quality titles in the AV recommendations engine. By encoding quality signals (metadata completeness, critical-reception proxies, engagement leading indicators) into the recommendation surface, the platform achieved a 22% increase in per-customer engagement. This is the RevAI archetype from the MuShuHaRi framework applied at pre-GenAI scale: intelligent signal → better surface → measurable revenue outcome.
Multi-team engineering at global scale
Building and managing multiple engineering teams simultaneously — each with distinct responsibilities across the combined tech stack supporting 70%+ of Amazon Video streams — required the same operational discipline that defines S+3 Agile at scale: shared KPIs, clear service boundaries, and a unified incident-response model. The global launch was not a single event; it was a rolling operation across time zones, regulatory environments, and content-licensing jurisdictions.
Key Results
| Metric | Outcome |
|---|---|
| Countries launched | 180+ |
| Share of total AV streams owned | 70%+ |
| Per-customer engagement lift | +22% (post-TQS deployment) |
| Algorithm innovation | First Title Quality Scoring platform at Amazon Prime Video |
| Platform continuity | Foundation carried into subsequent AV recommendation generations |
At global scale the recommendation engine is the product. Encode quality as a first-class signal and engagement follows — no feature ships more revenue than a surface that knows what good looks like.