> For the complete documentation index, see [llms.txt](https://docs.satisfilabs.com/resource-center/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.satisfilabs.com/resource-center/ai-agent-engine/reporting-and-analytics/analytics-dashboard/ai-agent-performance-report.md).

# AI Agent Performance Report

{% hint style="success" %}
**🚀 A New Analytics Experience Is Coming Soon**

We're evolving Analytics to provide deeper, more meaningful insights into your AI agent's performance. Over the coming weeks, you'll begin to see updates as we transition to a **conversation-based analytics experience**, designed to measure outcomes, not just messages.

The upgraded experience will feature:

* **Conversation-driven analytics** focused on user goals and outcomes
* **New visualizations and an improved UI** for easier reporting
* **Actionable insights** to better understand how your AI agent is performing
  {% endhint %}

### AI Agent Performance Report Overview

This report covers the AI agent's 30 day timeframe. 

To access AI agent analytics: 

1. Go to the [Agent Performance Console](https://agent.satis.fi/analytics/ai-agent)
2. Log in with the same credentials you use for the Satisfi Labs Dashboard.
3. In **Analytics**, click **AI Agent** 

<figure><img src="/files/Eksm9fSlfGxNX6iRGdJJ" alt=""><figcaption></figcaption></figure>

### **What you'll find inside, top to bottom:**

**Top KPIs**: the headline numbers: total messages and conversations handled, hours of staff time saved, and external link clicks the agent generated, each with a supporting detail (messages per conversation, share resolved without human escalation, share occurring outside business hours, and number of distinct URLs surfaced).

<figure><img src="/files/ciSvqZJ9SFQZ6DHGHT4g" alt=""><figcaption></figcaption></figure>

**Engagement Over Time**: when users actually talk to the agent. A daily trend of messages vs. conversations, an hour-by-hour view across the full 24-hour cycle, and a breakdown comparing weekday business-hours traffic to all other times, split by desktop and mobile.

<figure><img src="/files/cfeqPErp2zZlempo1cPF" alt=""><figcaption></figcaption></figure>

<figure><img src="/files/5E2OxKQbq6NPzcfoETyZ" alt=""><figcaption></figcaption></figure>

**Conversation Breakdown**: how people are reaching the agent and how it's responding. Message type (text, response buttons, quick replies), device channel (web, in-app, Facebook, email, SMS), and the split between generative (AI-written) and prewritten responses.

<figure><img src="/files/opMaox8nE4pEiiIr0ktH" alt=""><figcaption></figcaption></figure>

<mark style="color:$success;">**NEW:**</mark>**&#x20;Pop-up Deployment:** how the agent is surfaced across your web properties: total published pop-ups, and what share are set to open automatically on desktop vs. mobile.

<figure><img src="/files/TMkWIgxKjmOfNYLhTaG1" alt=""><figcaption></figcaption></figure>

**Topic Analysis**: what users are actually asking, and where the agent comes up short. The Top 10 Topics chart shows where message volume concentrates; the Topics Missing Coverage table shows which topics generate the most unanswered ("No Information Found") messages — your clearest signal of where to invest in new content.

<figure><img src="/files/WhbF2s1C5HS6O8qW88iJ" alt=""><figcaption></figcaption></figure>

{% hint style="info" %}
Every topic in **Missing Coverage** is clickable. Drill into a topic and you'll see its sub-topics broken out with their own unanswered counts, and from there, click any sub-topic to see the exact user messages that went unanswered, verbatim, ranked by how often they came up. It's a direct line from a high-level gap down to the real words a guest typed, so you know precisely what content to build next.
{% endhint %}

<figure><img src="/files/RGfkWg51iNrDJ7qrG2nE" alt=""><figcaption></figcaption></figure>

### Glossary

| Section    | Term                            | Meaning                                                                                                                 |
| ---------- | ------------------------------- | ----------------------------------------------------------------------------------------------------------------------- |
| Top KPIs   | Messages                        | Total number of distinct messages by end users. Shows all messages from end users to your AI agent.                     |
| Top KPIs   | Conversations                   | Total number of distinct conversations. A conversation is defined as all messages from a given end user on a given day. |
| Top KPIs   | Hours Saved                     | Estimated human-agent time saved = conversations × 5 minutes ÷ 60.                                                      |
| Top KPIs   | Linkouts                        | Clicks on links the AI surfaced that sent the guest to an external page                                                 |
| Top KPIs   | Published Pop-ups               | Chat pop-up widgets configured across the org's websites that are currently live.                                       |
| Top KPIs   | Auto-Open                       | Pop-ups set to open automatically (proactively surface the agent) on that device.                                       |
| Charts     | Generative vs Prewritten        | Share of AI responses that were LLM-written vs. scripted                                                                |
| Charts     | Conversations & Message Traffic | Daily trend of message volume vs. conversation count                                                                    |
| Charts     | Top 10 Resolved Topics          | Topics with the most AI-handled messages                                                                                |
| Charts     | Breakdown of Business Hours     | Desktop vs. mobile volume, split inside vs. outside business hours                                                      |
| Charts     | Traffic Segmentation Analysis   | How guests interact, by message type and device                                                                         |
| Charts     | Top Topics Missing Coverage     | Topics with the most unanswered messages — knowledge gaps                                                               |
| Drill-down | Sub-topics                      | A more specific breakdown within a topic; click to see the unanswered user messages behind it                           |
| Drill-down | Unanswered Messages             | Count of messages the AI couldn't answer for that topic                                                                 |
| Drill-down | User Input                      | The guest's exact wording for a question the AI answered with "No Information Found"                                    |

### FAQ

<details>

<summary>Will my historical data still be there when the transition is complete?</summary>

You will have access to 30-day back data. Your historical data can be accessed through **AI resolutions log** [in the Data section of the Dashboard](https://dashboard.satis.fi/Resource/Data) any time and will be available in the upgraded analytics that is coming soon.

</details>

<details>

<summary>Can I filter by date / endpoint?</summary>

The filters by date / endpoint / company name / language are not available in the report but will be in the upgraded analytics dashboard that is coming soon.

</details>

<details>

<summary>What's actually going to change?</summary>

You will have a brand new look at how we measure AI. The current platform scores individual messages, so it cannot assess the true quality of an agent interaction or demonstrate the business value Satisfi delivers.&#x20;

Our upgraded platform shifts focus away from outdated message-level metrics to provide a comprehensive, conversation-driven analytics experience, where each conversation is a set of tasks (user goals). And what we will measure is whether the user actually accomplished their goals or not. Also, you’ll have new analytics UI and visualization of your data, providing you a truly easy way to understand how your AI agent is performing.

</details>

<details>

<summary>My team can’t access the Console. Who can give access to it?</summary>

Email <support@satisfilabs.com> and let us know who we need to give access to.

</details>

<details>

<summary>I would like to view data greater than 30-days back from what this dashboard shows. Can I view that data in this dashboard or what is the best report to use?</summary>

This report gives you only a 30-day lookback. If you want to see data greater than 30-days back, you can use [**AI Resolutions Log**](https://dashboard.satis.fi/Resource/Data) report.

</details>