> ## Documentation Index
> Fetch the complete documentation index at: https://docs.roserx.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Activity & Analytics
> Track agent performance, trending topics, and user feedback from the RoseRx Activity page.
The Activity page is your central hub for understanding how your agents are performing. It provides real-time analytics across three key areas — performance metrics, content trends, and user feedback — so you can make data-driven decisions about your agent strategy.
Activity data is scoped to your current organization. Use the filter bar at the top of each tab to narrow results by initiative, agent, agent type, deployment, or time range.
## Accessing the Activity page
Navigate to **Activity** at the top of the left sidebar — it's also your home page when you sign in. You'll see a personalized greeting based on the time of day and three tabs for different analytics views.
***
The Performance tab gives you a high-level snapshot of your agent's usage and engagement metrics, with the ability to drill down into specific channels, regions, and devices.
### Filter bar
Use the filter bar at the top to refine the data you see:
* **Initiatives** — Filter by specific initiatives (country-level campaigns).
* **Agents** — Narrow to one or more specific agents.
* **Agent Types** — View data for a specific agent type, such as Patient Navigator or HCP Clinical Decision Support.
* **Deployments** — Narrow to specific deployments.
* **Time Range** — Select 7, 30, 60, or 90 days (30 days by default).
### Metric cards
Six key metrics are displayed as cards, each showing the current value and a **period-over-period percentage change** so you can spot trends at a glance. Click a card to display that metric on the chart below.
| Metric | What it measures |
| -------------------------- | ---------------------------------------------------------------- |
| **Visitors** | Unique users who interacted with your widget |
| **Conversations** | Total conversation sessions started |
| **Messages** | Combined count of all user and agent messages |
| **Msgs per Conversation** | Average number of messages per conversation session |
| **Avg. Conversation Time** | Mean duration of a conversation from first to last message |
| **AE Rate** | Percentage of conversations that triggered an adverse event flag |
A rising **Msgs per Conversation** metric often indicates strong engagement — users are finding the agent helpful enough to continue the conversation. A high **AE Rate**, on the other hand, may warrant a review of your adverse event sensitivity settings.
### Performance chart
Below the metric cards, the **Performance Overview** line chart visualizes the selected metric over time, with the previous period overlaid for easy trend comparison.
### Breakdown tables
Scroll down to see detailed breakdowns of your traffic across multiple dimensions. Each table shows conversation and message counts with period-over-period change:
| Breakdown | Details |
| ------------------- | -------------------------------------------------------------------------- |
| **Traffic Sources** | How users arrived — campaign attribution from UTM parameters and click IDs |
| **Channels** | Where conversations originated (e.g. website, direct link) |
| **Referrers** | Which external sites referred users to your widget |
| **Entry Paths** | The URL paths where users first encountered your widget |
| **Top Pages** | The most-visited pages where your widget is embedded |
| **Device Type** | Desktop, mobile, or tablet usage split |
| **Regions** | Geographic regions with country flags for easy identification |
| **Browser** | Browser usage breakdown (Chrome, Safari, Firefox, etc.) |
Use the **Regions** breakdown to understand where your audience is located. This is particularly useful for multi-market initiatives where you need to report engagement by country.
The Content tab helps you understand **what** your users are talking about. It surfaces trending topics and frequently used keywords to help you identify gaps in your knowledge base or emerging areas of interest.
### Topics trend chart
A **line chart** displays the top trending topics over time, showing how conversation themes shift day to day. Each topic is color-coded for easy comparison.
### Top topics table
Below the chart, a table lists all detected topics, sorted by frequency:
* **Topic name** — The topic label as configured in your agent's [Topics](/guides/agents/topics) settings.
* **Mention count** — How many times the topic was detected in conversations.
* **Trend direction** — Whether mentions are increasing, decreasing, or stable.
Click **View all** to expand the table and see the full list of topics.
### Top keywords table
A separate table shows the most frequently used keywords across all conversations, giving you a raw view of the language your users are using.
If you notice a keyword appearing frequently that isn't covered by an existing topic, consider creating a new topic for it. This helps you track emerging themes more precisely. See [Topics](/guides/agents/topics) for details.
The Feedback tab tracks user sentiment and satisfaction across your deployments. When users rate a response with a thumbs up or down, or leave a comment, it's captured here.
### Feedback overview
At the top, you'll see:
* **Total feedback count** — The total number of feedback responses received.
* **Feedback distribution** — A percentage bar showing the split between positive and negative feedback.
### Feedback over time
An **area chart** shows the volume of positive and negative feedback over the last 30 days, helping you spot patterns — for example, whether a recent knowledge base update improved sentiment.
### Feedback comments table
Below the chart, a table lists individual feedback entries with:
| Column | Details |
| ------------- | ---------------------------------------------------------- |
| **Time** | When the feedback was submitted |
| **Sentiment** | A color-coded badge — green for positive, red for negative |
| **Agent** | Which agent the feedback relates to |
| **Comment** | A preview of the user's comment |
**Click any row** to open the full details, including the complete comment text and the Session ID, which links to the conversation in [Conversations](/guides/monitoring/conversations).
Negative feedback is often the most valuable. Use it to identify specific areas where your agent's responses need improvement — whether that's a gap in the knowledge base, an overly cautious guardrail, or a tone mismatch.
***
## Tips for effective analytics use
Set a regular cadence — weekly is ideal — to review your Activity metrics. This helps you catch trends early and measure the impact of changes you've made.
The Performance Overview chart overlays the previous period automatically. Absolute numbers without context can be misleading.
Don't just collect feedback — act on it. Regularly review negative comments and use them to refine your agent's knowledge base and guardrails.
Use the Content tab to monitor how user interests evolve. This can inform your content strategy and highlight areas where your agent needs more knowledge.