> ## Documentation Index
> Fetch the complete documentation index at: https://docs.warrn.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Auto-grouping

> How Warrn rolls up similar alerts so you're not paging on the same thing five times.

When a flapping integration or one underlying problem creates several alerts that look basically the same, [Auto-Group Alerts](/components/teams/alert-intelligence#auto-group-alerts) rolls them into a single parent. You still see every alert; they just nest, so the list isn't full of duplicates.

## What grouping looks like

**On the parent alert**, a **Grouped Alerts** section lists the children with their severity, status, service, and team. A count badge on the parent (e.g. "3 grouped alerts") tells you at a glance how big the cluster is.

**On a child alert**, a header at the top says **"Grouped under \[Parent Name]"** with a similarity score badge:

* **95%+** - red. Near-duplicate.
* **90 - 95%** - amber. Strongly related.
* **Below 90%** - blue. Loosely related.

## Ungrouping

If Warrn grouped two alerts that shouldn't be together, click **Ungroup** on the child. The child becomes its own top-level alert again. The parent and other children are untouched.

This is reversible: if a future similar alert lands, it can be grouped under the same parent again.

## When grouping skips

Grouping only fires when:

* The team has **Auto-Group Alerts** turned on.
* A new alert is similar enough to an open alert (above the team's threshold).
* Both alerts belong to the same service.

Alerts that are too dissimilar, on different services, or arriving when no parent is open stay separate.

## Tuning the threshold

The slider on the team's [alert intelligence settings](/components/teams/alert-intelligence#auto-group-alerts) controls how strict grouping is. Higher numbers (95%+) group only near-duplicates; lower numbers (around 75%) catch looser similarities at the risk of over-grouping. Most teams leave the default at 85%.