Skip to content
English
  • There are no suggestions because the search field is empty.

What Is Anomaly Detection?

A Smart System for Monitoring Your Data

What is Anomaly Detection?

A smart system for monitoring your data

Anomaly Detection continuously monitors your key metrics and alerts you when something unexpected happens. There are two types of alerts — Anomaly Alerts and Custom Alerts — each designed for a different monitoring need.


Two types of alerts

Anomaly Alerts Uses machine learning to learn what "normal" looks like for your data and automatically detects unusual patterns. The system calculates a Z-score to measure how far current data deviates from your historical baseline. You control sensitivity using an Alert Threshold and filter out low-volume noise using a Metrics Threshold. Both conditions must be met for an alert to fire.

Custom Alerts Rule-based alerts where you define a fixed volume threshold for a specific tag or metric. If the volume crosses that number, an alert fires — no machine learning involved. Simple, predictable, and useful when you already know what number matters.


Alert Threshold vs Metrics Threshold

 

Anomaly Alerts

Custom Alerts

Alert Threshold

Low / Medium / High — controls ML sensitivity

Not applicable

Metrics Threshold

Minimum volume required to fire

The fixed number that triggers the alert


Alert frequencies

Both alert types support:

  • Hourly — checks data every hour and alerts immediately when a threshold is breached
  • Daily — analyses the previous day's data against a ~100-day historical baseline and sends a morning summary

What you see on the dashboard

The Alerts & Anomalies page shows all your recent and past alerts. You can switch between Anomaly Alerts and Custom Alerts using the tabs, filter by date range, and filter by metric and dimension.

Each alert card shows the tag name, volume, and an AI-generated summary of the tickets behind it.


Notifications

When an alert fires you receive an email and a Slack message to your designated channel. Both contain:

  • Tag name and metric
  • Current count, threshold, and delta
  • AI-generated root cause summary (Anomaly Alerts only)
  • 7-day and 30-day trend
  • Direct link to the source tickets

Feedback

Each anomaly card has a thumbs up and thumbs down option. Your feedback helps improve detection accuracy over time.


FAQ

What is the difference between Anomaly Alerts and Custom Alerts? Anomaly Alerts use machine learning to detect unexpected patterns automatically. Custom Alerts fire when a metric crosses a fixed number you define. Use Anomaly Alerts for intelligent monitoring and Custom Alerts when you have a specific threshold in mind.

What is the difference between Alert Threshold and Metrics Threshold? Alert Threshold controls ML sensitivity — Low catches minor changes, High only flags major ones. Metrics Threshold ensures only alerts with meaningful volume are sent. For Custom Alerts, the Metrics Threshold is simply the number you want to trigger on.

How do hourly and daily alerts differ? Hourly alerts fire as soon as a threshold is breached within an hour window. Daily alerts send a single morning summary based on the previous day's data.

How is the baseline calculated for Anomaly Alerts? The system calculates a Z-score from your historical data — roughly 100 days — to establish what normal looks like. Any significant deviation from that triggers an alert based on the sensitivity level you set.