Currently, the Alerts feature under Observability monitors two types of anomalies: usage anomalies and error anomalies (5xx status codes).
It would be great to extend error anomaly detection to also cover 404 spikes. A sudden increase in 404s can indicate broken links after a deployment, misconfigured rewrites/redirects, or SEO issues — all of which are worth catching early, just like 5xx errors.
Use Case
After deploying route changes, a spike in 404s would be the earliest signal that something went wrong, before users start reporting broken pages.
Proposal
Add an optional “4xx anomaly” (or at least “404 anomaly”) alert type alongside the existing error anomaly, using the same statistical detection approach.
Thanks Pauline! And yes, this feature request is 100% inspired by true events. We had a deployment very recently that generated a… let’s say generous amount of 404s. Nothing like real-world pain to fuel good feature requests.
We’ll look into Log Drains as a workaround, although it does look like a heavier setup!