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

# Home Dashboard

> Your account at a glance: conversations, messages, response time, leads and CSAT.

**Home** is the overview of what your agents have been doing. Pick a range — **This Month**, **7 Days**, **30 Days** or **90 Days** — and every number on the page follows it.

## The five metrics

<CardGroup cols={2}>
  <Card title="Conversations" icon="message">
    Distinct chat sessions in the range.
  </Card>

  <Card title="Messages" icon="messages">
    Total messages exchanged.
  </Card>

  <Card title="Avg response" icon="clock">
    How long your agent takes to reply, in seconds.
  </Card>

  <Card title="Leads" icon="user-plus">
    Contact details captured by the Lead Capture action.
  </Card>

  <Card title="CSAT" icon="face-smile">
    Share of rated conversations that were rated positively.
  </Card>
</CardGroup>

Each card shows the change against the previous period of the same length.

<Note>
  **A dash means no data, not zero.** If nothing was measurable in the range, the card shows **—**. A "0.0s" average response time would claim instant replies you never made, so AgntLabs does not print one.
</Note>

For **Avg response**, a rising number is bad news — that card turns red when it goes up, not green.

CSAT carries no change indicator at all, because a trustworthy previous-period figure is not available for it.

## What the numbers include

Home counts **every channel**, including your own test chats. A line under the metrics tells you what the range is made of — so if you have been testing an agent that is not deployed yet, you will see:

> Only internal test chats in this range — embed your agent to see customer data.

That is not an error. It is telling you nobody outside your account has talked to the agent yet.

## The trend chart

Conversations or messages over time — switch with the toggle above the chart.

## Resolution

**How conversations ended**, across five outcomes. AgntLabs works these out from the shape of the conversation when a session closes — there is no survey and no extra AI call.

| Outcome           | Meaning                                                                                     |
| ----------------- | ------------------------------------------------------------------------------------------- |
| **Auto-resolved** | The customer got what they came for — a real answer, or a completed action like a booking   |
| **Escalated**     | A human handoff happened; a support ticket was raised                                       |
| **Gap**           | An **on-topic** question your agent could not answer. This is the one to act on.            |
| **Deflected**     | An off-topic question, correctly turned away. Usually fine.                                 |
| **Abandoned**     | The agent asked a question and the customer never came back, or nothing meaningful happened |

<Warning>
  **Gap and Deflected are not the same thing and are deliberately kept apart.** Deflected means you correctly refused something outside your business. Gap means a real customer asked something reasonable and got nothing. Merging them would hide every real failure inside a healthy-looking number.
</Warning>

Only **closed** conversations are counted. A session is closed when the chat is closed, or after 12 hours of inactivity.

## Needs attention

A ranked list of the questions your agent could not answer, each with the reason:

* **No source covers this** — nothing relevant was found. Add a source.
* **Answer incomplete** — a source came back but did not contain the answer. Edit it.

Click one and the Knowledge base panel opens on the **Q\&A** tab with the question already filled in.

## Limit warnings

Two banners appear at the top of Home when you go over a plan limit:

* **Agent Limit Exceeded** — more published agents than your plan allows
* **Action Limit Exceeded** — more enabled actions than your plan allows

Each offers **Manage** (fix it) and **Upgrade** (raise the limit). Both can be dismissed.

## Home and Analytics do not always agree

They measure different things on purpose. [Analytics](/analytics/overview) excludes your own test chats so quality metrics are not skewed by your testing; Home includes them so you can see all activity. A young account can legitimately show a figure on one page and a dash on the other.