Metrics That Matter: Lessons from 100M Users

Beyond Vanity Metrics

When I started as an APM, I was obsessed with big numbers. MAU, downloads, page views — they felt important. They weren't.

The shift came when I started asking: what metric, if improved, would actually change user behavior?

For a streaming platform, that metric was Watch Time. Not views, not clicks — actual minutes spent watching content. Everything else was a leading indicator at best.

The Ratio Game

One of the most useful frameworks I developed was tracking ratios instead of absolutes:

Platform-Specific Nuance

Building for Android TV, Mobile Web, Desktop, and Jio phones simultaneously taught me that metrics need context.

A 5% Watch Time increase on Android TV might require redesigning the homepage banner. The same increase on Mobile Web might need a complete platform revamp. Same metric, wildly different investment.

Understanding the effort-to-impact ratio across platforms became essential for prioritization.

The Personalization Lever

When we introduced "My Corner" — a personalized content section with ML-generated recommendations — Watch Time increased 5%. That might sound small, but at scale, it represented millions of additional hours.

The insight wasn't just about personalization working. It was about where personalization works. Generic recommendations on the homepage didn't move the needle. A dedicated, curated space did.

Data as Compass, Not Destination

After two years of obsessing over metrics, I've learned they're a compass, not a destination. They point you in the right direction, but they don't tell you when you've arrived.

The best product decisions I've made came from combining quantitative signals with qualitative understanding. Numbers tell you what's happening. Talking to users tells you why.