> For the complete documentation index, see [llms.txt](https://docs.crobox.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.crobox.com/how-to-guides/sagent/monitoring/overview.md).

# Overview

Monitor Sagent health trends with widgets, locale filters, and time ranges.

Use **Overview** to monitor your Sagent's health over time. It helps you spot quality drops, tool issues, and unusual conversation patterns.

### How to use Overview

{% stepper %}
{% step %}

#### Open Monitoring for a single Sagent

Open your Sagent and click **Monitoring** in the top navigation.
{% endstep %}

{% step %}

#### Choose your locale and time range

Use the **Locale** filter to focus on one market or language. The default is **Global - All**.

Use the date picker to define the period you want to review.

Start broad to identify trends. Use shorter ranges to inspect recent changes.
{% endstep %}

{% step %}

#### Review the four monitoring widgets

Scan all four cards together. They explain different parts of Sagent health.

Use them as a connected view, not isolated charts.
{% endstep %}

{% step %}

#### Use changes as context

Look for drops after recent updates.

The trend matters more than a single point.
{% endstep %}

{% step %}

#### Move to conversation review

Use **Overview** to spot patterns first.

Then use **Conversations** to inspect the underlying sessions.
{% endstep %}
{% endstepper %}

### Monitoring widgets

#### Conversation Health

This widget shows week-over-week score changes by quality dimension. It uses a line chart over time.

Use it to spot stable, improving, or declining quality. Dips often follow prompt or catalogue changes.

The score is a quality signal from monitored conversations. It is not a shopper satisfaction score or purchase outcome. Compare periods with similar locale coverage and enough conversations.

#### Average Tool Response Time

This widget shows response-time trends for each tool. It uses a line chart over time.

Use it as a health check for latency issues. Watch for sudden dips or spikes after deployments.

The metric comes from tool activity during Sagent conversations. It measures tool responsiveness, not the full shopper wait time. No data can mean no eligible tool activity occurred.

#### Flagged Conversations

This widget shows flagged conversations over time. It uses a stacked bar chart by flag type.

Use it to identify increasing risk or off-path behavior.

Flags come from monitoring classifications. A flag identifies a review signal. It does not prove that Sagent gave an incorrect answer.

Current flag categories are:

* Guardrail trigger
* Customer support
* Small talk

#### Average Tool Usage

This widget shows how often each tool succeeds in conversations. It uses a line chart per tool.

Use it to catch unexpected drops in tool usage. This can show when part of the experience is used less.

The metric comes from successful tool use in selected conversations. A lower average can reflect changed shopper needs. It does not necessarily indicate tool failure.

### Read trends effectively

Start with these checks:

1. Compare the current period with the previous one.
2. Look for changes across more than one widget.
3. Match the shift to a product or configuration change.

For example:

* **Conversation Health drops** and **Flagged Conversations rise** — review recent updates and conversation patterns.
* **Tool Response Time worsens** and **Tool Usage drops** — check whether response speed affects the experience.
* **Tool Usage drops** with stable response time — review whether the conversation flow changed.

### Filters and defaults

Monitoring includes two primary filters:

* **Locale**
* **Time range**

The default locale is **Global - All**.

Use the same date range and locale when comparing widgets.

### Best practices

Review Monitoring regularly, not only when something breaks. Weekly reviews work well for trend tracking.

Use it after every major Sagent update.

When you see a change, compare:

* Locale
* Time range
* Recent Sagent updates


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:

```
GET https://docs.crobox.com/how-to-guides/sagent/monitoring/overview.md?ask=<question>&goal=<endgoal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.