Most systems today are over-instrumented and under-observed. You’ve got a dozen dashboards and a graveyard of alerts, but when things break, you’re still asking: “What just happened—and did users notice?”

This talk is about fixing that, starting with a simple idea: Service Level Indicators (SLIs) are not just metrics—they’re contracts between your system and your users. We’ll walk through building SLIs that actually reflect user experience, instrumenting them using OpenTelemetry, and wiring them into Azure’s Health Model to create a platform-native reliability feedback loop.

You’ll learn how to define meaningful indicators (not just what’s easy to measure), surface them in Azure Monitor, and use them to drive Service Level Objectives (SLOs), error budgets, and deployment gates. Along the way, we’ll discuss the anti-patterns that kill signal quality, how to avoid “observability theater,” and why developers—not just SREs—should own reliability from day one.

Expect strong opinions, live demos, and answers to questions most dashboards can’t.

Discover how to build meaningful Service Level Indicators (SLIs) using OpenTelemetry and Azure’s Health Model. This session dives into practical instrumentation, turning raw telemetry into actionable reliability signals, and integrating them into real-world SLOs and error budgets. Designed for developers who want to own reliability and move beyond noisy metrics to signals that truly matter.