AppSignal has launched log-based metrics in beta as part of AppSignal Labs. The feature lets developers extract counters, gauges, and measurements from any log source at ingestion time — without storing every log line. Three metric types are supported: Count (matching log lines), Gauge (attribute value at ingestion), and Measurement (histogram of numeric attributes). A key pattern is placing metric actions before filter actions so you can track error rates or latency from high-volume logs while discarding the noise. Metrics integrate with AppSignal dashboards, anomaly detection, alerts, and the metrics API. The UI includes a forms-based editor in Logs Explorer with live preview, per-line metric creation from numeric attributes, and a dedicated Metrics management page. No backfill against historical logs is available.
Table of contents
How log-based metrics workWhy this changes what you keepA few concrete examplesWhat's actually newWhat "in beta" means for this oneHow to try itRelated readingSort: