Pulse: Anomaly detection that respects your attention

Alerting tools fail in one consistent direction: too much noise. Most systems treat every deviation as signal, which means teams learn to ignore alerts within weeks of turning them on. A 3% drop in daily active users on a Sunday morning is not the same as a 3% drop on the first day of your biggest marketing campaign. Pulse was designed around that distinction. It builds an adaptive baseline for each metric using your business’s own historical patterns, and accounts for day-of-week, seasonal, and growth-curve variation before deciding whether something is worth your attention.

How severity scoring works

Severity scoring adds a second layer of filtering. When Pulse does fire, it attaches a severity score based on how far outside the baseline the deviation sits, how long it has persisted, and whether correlated metrics are also moving. A single metric dipping briefly scores low. Multiple related metrics moving together in an unexpected direction scores high. That scoring feeds directly into routing logic, so a low-severity alert goes to a Slack channel and a high-severity alert can page on-call.

Getting started

Pulse is available on Pro and Enterprise plans. Setup requires no manual threshold configuration. Connect your metrics, let Pulse observe for 72 hours to establish baselines, and it handles the rest. Routing destinations, severity thresholds, and suppression windows can all be adjusted from the Pulse settings panel once baselines are active.