Anomaly detection
Last updated 2026-06-14
Definition
Anomaly detection is how Quri decides a metric has moved enough to be worth your attention. It learns each metric’s recent baseline, then flags the points that spike or drop well outside that normal range. That is the signal behind every proactive alert — a flagged change, not just a raw number.
How to do this in Quri
- Send Quri a metric series by connecting a source or pasting numbers into a free tool.
- Quri builds a trailing baseline from the recent points in that series.
- Each new point is compared to the baseline to see how far it strays.
- Points outside the normal range are flagged as spikes or drops for you to review.
Frequently asked
- What counts as an anomaly?
- A point that sits far enough from the metric’s recent baseline to be unlikely by chance. Quri grades the distance so a small wobble is ignored and a real spike or drop is flagged.
- Can I try anomaly detection without connecting anything?
- Yes. The free Anomaly Spotter tool lets you paste a series and see which days Quri would flag, with no signup.