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

# Evaluations

> Review and assess the quality of escalated conversations handled by your live-chat team.

**Evaluations** (previously called *Escalation Insights*) lets you analyze the quality and outcomes of conversations that were escalated from the AI agent to human operators. Use this view to understand how your live-chat team is performing and where coaching or process improvements are needed.

<Note>
  Evaluations is a **Beta** feature. Metrics and scoring criteria may evolve as the feature matures.
</Note>

## What you'll see

The Evaluations view surfaces quality metrics for escalated conversations, helping you answer questions like:

* How quickly are escalated conversations being picked up?
* Are customers satisfied after speaking with a human agent?
* Which operators are handling the most escalations?
* What topics are most commonly escalated?

## Filters

Use the filter bar to narrow down the data:

* **Date range** — select a period or use presets
* **Channel** — filter by communication channel
* **Operator** — filter by the team member who handled the conversation

## Common use cases

* **Quality assurance** — review a sample of escalated conversations to ensure operators follow best practices
* **Training** — identify patterns where operators struggle and use real conversations as coaching material
* **Process improvement** — spot recurring escalation reasons that could be resolved by updating the AI agent's knowledge or workflows

<Tip>
  Combine Evaluations with the **Escalations Board** for a complete picture: use the board for real-time triage and Evaluations for after-the-fact quality review.
</Tip>