Sentry expands from errors to APM
Sentry: the $128M ARR error tracker disrupting New Relic
This shift is really about turning a narrow debugging tool into a much larger engineering system of record. Error tracking tells a team where code broke. APM tells them what slowed down, which service caused it, which users were affected, and what happened right before and after. By adding tracing, session replay, and code coverage on top of the same SDK already embedded in apps, Sentry can move from helping one developer fix one bug to helping whole engineering teams monitor live software in production.
-
Sentry entered through developers, not central IT budgets. Its SDK sits inside app code, auto captures crashes and context, and spreads bottom up. That made error tracking cheap to adopt and sticky. APM is the natural expansion step because the same instrumentation can also collect latency, traces, and user session data.
-
Datadog and New Relic sell broader observability suites to larger engineering organizations. Datadog prices APM by host and layers on adjacent products like session replay and error tracking. New Relic bundles APM, tracing, error tracking, logs, and digital experience tools into a full platform. Sentry is attacking that spend with a simpler developer first workflow.
-
The revenue logic is much bigger than fixing bugs faster. Sentry’s core error tracking customers historically spent more like a self serve developer tool, while incumbent APM vendors often sell into organization wide budgets with far higher contract values. That is why winning APM matters, it lifts both average revenue per customer and the size of market Sentry can sell into.
The next phase of observability will be won by the vendor that makes root cause analysis feel automatic inside the developer workflow. Sentry is moving in that direction by expanding from errors into performance, replay, and code level context. If it keeps converting self serve adoption into broader monitoring spend, it can become the default front door for application observability, not just error triage.