Knowledge Agent Performance Overview (Website Channel)
Introduction: Understanding Agent Performance
Once your Knowledge Agent is live and interacting with users, the Agent Overview page becomes your central command center for understanding how it is performing.
Important: Performance metrics are tracked at the agent level, not at the Screen or Engagement level. This means you can see exactly how each individual agent is contributing to your goals, independent of the building blocks it uses.
This guide covers the performance dashboard specifically for Website Knowledge Agents.
Accessing Agent Overview
From the main navigation sidebar, click
Agents.In the Agents list, click on the title of the agent you want to review.
You will land on the Overview tab.
Overview Dashboard Layout
The Overview page is divided into several key sections.
1. Key Metrics (Top Cards)
At the top of the page, you will see five core metrics that give you an immediate snapshot of agent performance.
Metric | Description | What It Tells You |
|---|---|---|
Screen Loads | Number of times the agent widget loaded on a page | Reach – how many pages showed the agent |
Screen Impressions | Number of times users actually saw the agent (scrolled into view) | Visibility – how many users were exposed |
No. of Questions asked | Total number of questions users submitted to the agent | Engagement – how many users actively interacted |
Avg Time Spent | Average time users spent interacting with the agent | Depth – how engaged users were with responses |
Conversions | Number of times users completed a CTA (email form, link click, etc.) | Value – how many leads or actions were generated |
Time Period Filter:
Use the dropdown next to "Key metrics from:" to switch between time periods:
Past Week
Past Month (default)
Past 3 Months
Custom Range
2. Live Preview
Before launching your agent live, you can preview how it will appear and behave on an actual page.
Location: Found near the metrics section or within the agent information header
Label: "Your Live preview will be displayed from this article:" followed by a URL
How to Use Live Preview:
Step | Action |
|---|---|
1 | Enter a URL where you want to test the agent |
2 | The system will display a live preview of how the agent will appear on that page |
3 | Interact with the preview to test questions, responses, and CTAs |
4 | Make adjustments to your agent configuration as needed |
5 | Once satisfied, launch the agent live |
Use Case:
Test placement and styling on different page types before going live
Verify that CTAs and Engagements appear correctly
Ensure the agent pulls relevant content from your Collection
Debug any issues in a safe environment without affecting real users
Note: The Live Preview mode does not track metrics or capture leads. It is purely for testing and validation before launch.
3. Agent Information Header
Below the metrics, you will find basic information about the agent:
Field | Description |
|---|---|
Agent Name | The name you assigned during creation |
Created at | Date and time of creation |
Created by | Email of the user who created it |
View all activities | Link to see detailed activity logs |
4. Timeline Chart
The timeline chart visualizes performance trends over the selected time period.
X-axis: Dates (e.g., 13-Apr-26 to 13-May-26)
Y-axis: Metric values (e.g., 0 to 120+)
What you can track:
Total Screen Impressions over time (default view)
Use the dropdown to toggle between different metrics:
Screen Loads
Screen Impressions
Questions Asked
Conversions
Use Case: Identify spikes or dips in performance. For example, a spike in questions might correlate with a new blog post or marketing campaign.
5. Action Buttons
Button | Function |
|---|---|
Refresh Agent | Manually refresh the agent's knowledge or cache |
Refresh FAQs | Refresh the AI-generated FAQs and clickable questions on the Knowledge Agent. Note: If FAQs were added manually (not AI-generated), this button has no effect. |
Edit Agent | Go to the agent editor to modify settings |
Agent Activities | Shows a timeline of recent changes or updates made to the agent, including who made the change and when |
6. Settings Panel (Right Sidebar)
The Settings panel displays the current configuration of your agent for quick reference.
Overall Settings
Setting | Description |
|---|---|
Agent Type | Knowledge Agent, Polling Agent, etc. |
AI dynamic engagement | On/Off – Whether AI generates dynamic content |
Dark theme | On/Off – Visual theme setting |
Primary color | Brand color code (e.g., |
Language | Selected language (e.g., English) |
Journey ID | Unique identifier for the agent |
Domain Settings
Setting | Description |
|---|---|
Type | Full Domain, Section of Domain, Single Page, Regex |
Website | The domain where the agent runs |
Sections | Specific URL paths where the agent is active |
Exceptions | URLs where the agent is NOT active |
Deeper Analytics Tabs
Beyond the overview, you can dive into more detailed analytics using the additional tabs.
7. Reference Clicks Tab
This tab shows which source links or references users clicked on within the agent's responses.
Column | Description |
|---|---|
Page Title | The title of the referenced page |
Page URL | The link to the referenced page |
Click Count | Number of times this reference was clicked |
Use Case: Understand which content is most valuable to users. High-click references indicate helpful answers that users want to explore further.
8. Feedback Tab
This tab shows user ratings and comments on the agent's responses.
Column | Description |
|---|---|
Page Title | The page where feedback was given |
Feedback Type | Star Rating, Text Comment, etc. |
Rating | Star rating (1-5 stars) |
Comment | Any text feedback provided by the user |
Use Case: Identify responses that need improvement. Low-rated responses with specific comments (e.g., "too informal") can be used for fine-tuning your Engine.
9. Page Analytics Tab
This tab provides page-level breakdown of agent performance.
Column | Description |
|---|---|
Page Title | The specific webpage where the agent appeared |
Screen Impressions | How many times the agent was seen on that page |
Engagement Count | How many interactions occurred |
Screen Load | How many times the agent loaded on that page |
Avg Time Spent | Average interaction time on that page |
Use Case: Identify which pages drive the most engagement. Optimize underperforming pages by reviewing content or placement.
10. Leads Tab
This tab displays all captured lead information from users who completed CTAs attached to this specific agent.
Note: Leads shown here are only those captured by CTAs added to this agent. CTAs attached to other agents will not appear in this view.
Lead Identification
Type | Definition |
|---|---|
Identified Profiles | Profiles that have at least a phone number OR an email address |
Anonymous Profiles | Profiles with no phone or email (e.g., only form fields like company name) |
Filters Available
Filter | Description |
|---|---|
Identified / Anonymous | Toggle between identified and anonymous profiles |
CTA Filter | Filter leads by the specific CTA they completed |
Dynamic Field Filters | Filter by any custom field collected in the form (company, job title, etc.) |
Lead Table Columns
Column | Description |
|---|---|
User's email address (if captured) | |
Phone | User's phone number (if captured) |
Lead Details | Any additional dynamic information captured (company name, job title, custom answers, etc.) |
Date Collected | When the lead was captured |
What are Lead Details?
Lead Details contain any extra information captured beyond email and phone. This depends entirely on the dynamic form the user filled out. Examples include:
Company name
Job title / role
Industry
Custom survey answers
Any other fields defined in your CTA form
Use Case: Export or review leads generated by this agent. Filter by specific CTAs to see which offers drive the best leads. Use dynamic field filters to segment by company size, role, or any other custom data.
Troubleshooting & Common Scenarios
Scenario | What It Means | Action |
|---|---|---|
High Screen Loads but low Questions Asked | Users see the agent but don't interact | Review placement, CTA visibility, or question prompts |
Low Avg Time Spent | Users are leaving quickly | Check response quality or relevance of answers |
High Conversions but low Impressions | CTA is very effective but reach is limited | Expand targeting to more pages or sections |
Page Analytics show zero engagement on certain pages | Agent may not be placed correctly on those pages | Review placement settings for that page type |
Refresh FAQ button does nothing | FAQs were added manually, not AI-generated | No action needed; button only works for AI-generated FAQs |
Live Preview not showing agent | URL may not match targeting rules | Check that the preview URL is covered by your domain/section settings |
Best Practices for Monitoring Performance
1. Preview Before Launching
Always use Live Preview on 2-3 representative URLs before launching an agent. Verify placement, styling, and response quality.
2. Review Weekly
Check key metrics weekly to catch trends early. Sudden drops in impressions may indicate placement issues.
3. Compare Periods
Use the time filter to compare month-over-month or week-over-week performance.
4. Analyze by Page
Use Page Analytics to identify top-performing and underperforming pages. Optimize placement on low-performing pages.
5. Track Lead Quality by CTA
Use the CTA filter in the Leads tab to see which CTAs generate the most (and best) leads. Double down on what works.
6. Use Reference Clicks to Improve Content
High click-through rates on specific references suggest users find that content valuable. Consider promoting it more prominently.
7. Act on Text Feedback
Low-star ratings with specific comments are gold. Use them for Engine fine-tuning.