Anomalies in Insight

Summary

In Insight, open the Anomalies tab. It compares the same metrics across agents for the selected period and helps reveal values that differ most from the average.

An anomaly does not mean that an indicator became worse over time. It only means that the result stands out from other agents in the current selection.

When to use it

• you need to understand which agent behaves differently from the rest;
• an indicator looks unusual, but the cause is not yet clear;
• the AI summary highlighted an issue and you need to check whether it is local or widespread.

How to read the result

1. Select the Anomalies tab.
2. In problem metrics, find the indicators with the most noticeable deviations.
3. In agents requiring attention, check where those deviations are concentrated.

Open a metric or agent card to see the details. Do not judge by the percentage sign alone: an increase is better for some metrics, while a decrease is better for others.

If there is not enough data

The comparison requires at least two agents with the same metric in the selected period. If there is nothing to compare, increase the period, loosen the filters, or check whether other agents collect this metric.