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

# Outcome dashboard: track one agent's results

> Build a dashboard of charts for one agent from its calls and its own outcome fields, compare periods, and read the Clarity and Observability tabs.

> [!NOTE]
>
> The Outcome dashboard is in beta and switched on for selected projects.

The Outcome dashboard is a page of charts for one agent. It starts with charts of call volume, resolution, sentiment and transfers, and you can add your own from the fields your agent fills in after each call. Two more tabs show where the agent lacks knowledge (**Clarity**) and where calls fail (**Observability**).

![The Outcome dashboard for Order Support Flow, Last 90 days compared with the previous period: Total Calls, Avg Call Duration, Avg AI Latency and the Call Volume trend](https://mintcdn.com/hamsa-452fa79a/2654KIaxnTY04VOj/images/agents/monitor/outcome-dashboard-light.webp?fit=max\&auto=format\&n=2654KIaxnTY04VOj\&q=85\&s=9f8633a31971f352467a77fcc9bab4a6)

## Before you start

* Viewing needs **View** on **Outcome dashboard** ("See the outcome dashboard."). Changing it needs **Edit** ("Change what the outcome dashboard tracks."). Owners and Admins have both by default; Users have neither. See [Roles and permissions](https://docs.tryhamsa.com/overview/projects/roles-and-permissions.md).
* To chart your own fields, the agent needs outcome fields in its **Outcome** settings, and it must be published. Fields come from the published agent.

## Open the dashboard

In the agents list, click the chart icon in the agent's **Outcome** column ("How this agent's calls went: resolved, transferred, sentiment and more."), or open the agent's row menu and choose **Outcome dashboard**.

The header shows how many calls fall in the period, for example "1,240 calls in selected period".

## Choose the period and compare

* **Period**: **Last 7 days**, **Last 30 days**, **Last 90 days** (the default), **This week**, **This month** or **Custom range**. A range can be up to 1000 days.
* **Compare**: **No comparison**, **Previous period** (the same number of days just before), **Year ago** (the same days a year earlier) or **Custom range**. Charts then show the earlier period beside the current one.

Click a part of a chart (a donut slice, a bar) to filter the dashboard by it. Filters show as chips after "Filtered by:"; **Clear all** removes them.

## The three tabs

### Analytics

Charts of this agent's calls. It starts with these widgets:

| Widget | What it shows |
| - | - |
| **Total Calls** | Calls in the period. |
| **Avg Call Duration** | Average call length. |
| **Avg AI Latency** | Average time the AI took to respond. |
| **Answer Rate** | Share of calls that were picked up. |
| **AI Success Rate** | Share of calls the AI resolved. |
| **Transfer Rate** | Share of calls transferred to a person. |
| **Call Volume** | Calls per day. |
| **Avg Call Duration Trend** | Average call length per day. |
| **AI Latency Trend** | Average AI response time per day. |
| **AI Resolution** | Resolved and not resolved calls, as a donut. |
| **Caller Sentiment** | Positive, neutral and negative calls. |
| **Call Direction** | Inbound and outbound calls. |
| **Transferred to Human** | Transferred and not transferred calls. |

![Further down the Analytics tab: the AI Resolution, Caller Sentiment, Call Direction and Transferred to Human donuts, then Answer Rate, AI Success Rate and Transfer Rate](https://mintcdn.com/hamsa-452fa79a/2654KIaxnTY04VOj/images/agents/monitor/outcome-dashboard-donuts-light.webp?fit=max\&auto=format\&n=2654KIaxnTY04VOj\&q=85\&s=1f2061f7cba5811fdfd0297f9882e2ad)

These widgets, and the ones you add, count calls with the status Completed or Forwarded, so calls transferred to a person are included. **Answer Rate** also counts calls that ended as No answer or Terminated, so that unanswered calls are part of the rate. Resolution, transfers and sentiment come from the AI analysis of each call. See [How calls are scored](https://docs.tryhamsa.com/agents/monitor/analytics.md#how-calls-are-scored).

You can add your own widgets here (see below). **Manage widgets** shows or hides the starting widgets.

### Clarity

Where the agent couldn't help, over completed and forwarded calls. These widgets are fixed, and each shows only when the agent's calls can answer it.

| Widget | What it shows |
| - | - |
| **Fully Handled** | Calls resolved without a person, out of the calls the analysis judged. |
| **Escalated / Deflected** | Calls handed to a person, out of the calls the analysis judged. |
| **Top User Keywords** | The topics callers asked about most that the agent couldn't handle. |
| **Topic Coverage Gaps** | The same topics, as shares of a donut. |
| **Knowledge Gap Reasons** | Why the agent couldn't answer: missing from the knowledge base, an outside system didn't have the data, or the agent didn't find what the knowledge base has. |
| **Knowledge Gap Trend** | Knowledge gaps over time. |

![The Clarity tab: Fully Handled, Escalated / Deflected, Top User Keywords and Topic Coverage Gaps](https://mintcdn.com/hamsa-452fa79a/2654KIaxnTY04VOj/images/agents/monitor/outcome-dashboard-clarity-light.webp?fit=max\&auto=format\&n=2654KIaxnTY04VOj\&q=85\&s=f6e4f92a69a3b6255d409fe818e32434)

The topics are the ones listed under **Asked outside its scope** in each call's [call analysis](https://docs.tryhamsa.com/agents/monitor/call-details.md#call-analysis). Use them to decide what to add to the [Knowledge base](https://docs.tryhamsa.com/agents/knowledge-base/overview.md) or the prompt.

### Observability

Where calls fail, over completed, forwarded and failed calls. These widgets are fixed.

| Widget | What it shows |
| - | - |
| **Call Success Rate** | Calls the AI resolved, out of the calls the analysis judged plus the calls that failed. |
| **Successful Calls** | The number of calls the AI resolved. |
| **Avg Latency** | Average time taken to process a response. |
| **Daily Success Rate Trend** | Call success rate per day. |
| **Peak Failure Distribution** | Failure types per day. |
| **Error Breakdown** | Errors by type: speech to text, model, speech, and failed calls. A call with two kinds of error counts in both. |
| **Failed Calls** | A table of calls that didn't complete, up to 50 rows. |

A widget whose metric the agent doesn't report shows "Not integrated".

## Add a widget

**1. Open the dialog**

On the **Analytics** tab, click **Add widget**. On an empty dashboard, click **Add your first widget**.

![The Add widget dialog with Title, Description, Graph type, Field, Compare by, Group by and Filters on the left and the widget preview on the right](https://mintcdn.com/hamsa-452fa79a/2654KIaxnTY04VOj/images/agents/monitor/outcome-dashboard-add-widget-light.webp?fit=max\&auto=format\&n=2654KIaxnTY04VOj\&q=85\&s=200690a561442996851463eadab6a68c)

**2. Pick a graph type**

Under **Graph type**, choose **Trend** (the default), **Bar**, **Donut**, **KPI**, **Table**, **Stacked bar**, **Radar** or **Heatmap**.

**3. Name it**

Enter a **Title**, and a **Description** if you like.

**4. Choose what to chart**

Pick a **Field**: one of your agent's outcome fields, or a platform field such as duration or sentiment. For a field that holds several keys, pick a **Sub-field**. Then pick a **Measure** (see below). For a trend, pick **Group by**: **Day**, **Week** or **Month**.

**5. Narrow it down (optional)**

Use **Compare by** to split the chart by another field (a stacked bar needs one) and **Show only** to keep some values. Add **Filters**: pick values **From your calls**, or type one (it matches the exact text, capital letters included). Set **Comparison direction** to say whether higher or lower is better, which colors the change when you compare periods. Set a **Custom date range** to give this widget its own dates.

**6. Check and save**

The **Widget preview** shows the chart as you go. Click **Add widget** to save it.

A new outcome field can take a few minutes to show in **Field** after you publish the agent.

### Measures

| Measure | What it calculates |
| - | - |
| **Total call count** | How many calls. |
| **Count per value** | How many calls have each value of the field. |
| **Average value** | The average of a number field. |
| **Percentage distribution** | Each value's share of calls. |
| **Count total items** | How many items a list field holds, across calls. |
| **Values frequency** | How often each item in a list field appears. |
| **Cross-tabulation** | Counts for each pair of values of two fields. |
| **Average per sub-field** | The average of each key inside a field. |
| **Rate (% of total calls)** | The share of all calls that match the widget's filters. |

## Change the layout

Click **Edit layout**, then drag widgets to reorder (**Drag to reorder**) or resize (**Drag to resize**). Click **Done editing** when you're finished.

Each widget's **Widget options** menu has:

* **Expand**: the widget's data as a table.
* **Edit**, **Duplicate** and **Delete** (for your own widgets), or **Hide** (for starting widgets).
* **Export CSV** and **Export Excel**: the widget's data.

If you change the dashboard's period while some widgets have their own dates, you're asked **Reset custom dates?**: **Keep custom dates** or **Reset & apply**.

The layout is saved for the agent and shared by everyone in the project: when you change it, your teammates see the change. If saving fails, you see "Couldn't save the layout. Your change was undone."

## Good to know

* Members who can view but not edit see the dashboard as it is. On an empty dashboard they see "No widgets yet".
* Widget titles you see in English are the platform's own names. In other languages they are translated.
* The dashboard and [Analytics](https://docs.tryhamsa.com/agents/monitor/analytics.md) both count completed and forwarded calls for resolution and transfers. Analytics also counts calls in other statuses in some of its totals, so a total can differ between the two for the same agent and dates.

## Related

> [card] [Analytics](https://docs.tryhamsa.com/agents/monitor/analytics.md)
>
> Figures across all your agents, and how calls are scored.

> [card] [Call details](https://docs.tryhamsa.com/agents/monitor/call-details.md)
>
> One call's analysis and outcome fields.

> [card] [Knowledge base](https://docs.tryhamsa.com/agents/knowledge-base/overview.md)
>
> Fill the gaps Clarity finds.

> [card] [Conversation log](https://docs.tryhamsa.com/agents/monitor/conversation-log.md)
>
> The calls behind the charts.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.