> For the complete documentation index, see [llms.txt](https://presight.gitbook.io/presight-v1.0/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://presight.gitbook.io/presight-v1.0/presight-iq-beta/auto-insights-beta/insight-type-2-based-on-strangeness.md).

# Insight Type 2 - Based on Strangeness

The second type of insight PresightIQ suggests to you is the direct sub-segment of the metric whose **behavior is the most unusual/strange**.

For example:

> <mark style="color:green;">**Sales of Region Europe**</mark> typically fluctuates in the **80-120** range; and it was expected to be <mark style="color:blue;">**100**</mark> for **March 2023**. When the numbers came back, it was actually <mark style="color:red;">**80**</mark>.&#x20;
>
> <mark style="color:green;">**Sales of Product A**</mark> typically fluctuates in the **40-60** range; and it was expected to be <mark style="color:blue;">**50**</mark> for **March 2023**. When the numbers came back, it was actually <mark style="color:red;">**30**</mark>.&#x20;
>
> Even though they share the same gap (20 short), the one of **Product A Sales** was **much more unusual** given its typical trend.
>
> Presight will recommend **Sales of Product A** to you for the second type of insight.

There are different levels of strangeness for the insights Presight recommends to you:

* <mark style="color:green;">**Expected**</mark>
* <mark style="color:yellow;">**Quite Strange**</mark>
* <mark style="color:red;">**Highly Strange**</mark>

{% hint style="warning" %}
To make sure that all of our recommendations are meaningful to your diagnosis of the anomaly, all segments need to pass a certain threshold of significance in value. So when we say "biggest", think biggest of all the significant enough segments.
{% endhint %}