Lemwatch compares each endpoint's current response time against a rolling 24-hour baseline and alerts when it degrades by 2× or more.
Each endpoint keeps a 24-hour rolling average response time. If the current p50 goes above 2× baseline for 3 consecutive checks, we open a degradation incident.
APIs have wildly different normal ranges. A 500 ms baseline API paged at 500 ms is fine; a 50 ms baseline API at 500 ms is broken. Baseline math catches both.
Site → API Endpoints → <endpoint> → Alerts.
Yes — the "max response time" field overrides baseline logic.
24 hours of data. Until then only status-code alerts fire.