SLOs for Public APIs: Start With Two Metrics

Observability tools tempt you to chart everything. For public APIs, start with availability and p95 latency on your top three routes. Define an error budget in plain language: how many bad minutes per month you can spend.

Synthetic checks should hit the same routes your partners use in sandboxes, not only health endpoints that return 200 with empty bodies.

Alert when you burn budget fast, not when a single spike happens during deploys. We template alerts in the API Observability Bootcamp so learners leave with copy-paste YAML and a one-page exec summary.

Remember: SLOs are agreements with stakeholders. If product expects five-nines on a beta route, write that down and negotiate — do not hide mismatch in dashboard granularity.