User Behaviour Analysis
Get AI-powered recommendations for improving page engagement—combining scroll behaviour, click patterns, time spent, and meaningful engagement data with visual analysis to identify exactly what to change and where.
Quick start
1. Select the page to analyse from the Pages dropdown
2. Set filters: Date Range (30 days recommended), Device, and optionally Referrer
3. Click Analyse with AI
4. Review the recommendations—each includes priority, location, and specific changes
5. Click Add to Kanban on any recommendation to create a task
What the analysis provides
The AI analyses your page screenshot alongside behavioural data to deliver actionable recommendations. The analysis combines multiple data sources that would take hours to correlate manually.
Data sources used
| Source | What it reveals |
| Click patterns | Where visitors interact and what they ignore |
| Scroll depth | How far visitors travel down the page |
| Time spent | Which sections hold attention and which are skipped |
| Meaningful engagement | Zones where visitors pause for 2+ seconds (deliberate attention) |
| Page screenshot | Visual context for understanding what content exists at each depth |
Understanding the output
The analysis appears below the dashboard charts with three main sections:
Page Optimisation Recommendations
The primary output—specific, actionable changes organised by scroll zone. Each recommendation includes:
| Field | Description |
| Scroll Zone | The 5% depth band where the change should be made (e.g., 15–20%) |
| Title | Brief headline summarising the recommendation |
| Description | One-sentence summary of what to do |
| Copy/UX Suggestion | Exact wording or strategic direction for the change |
| Implementation Specs | Measurable details: sizes, spacing, positions, contrast ratios |
| Current State | What exists now at this scroll depth |
| Recommended Change | Detailed implementation guidance |
| Justification | Why this change matters—references specific metrics |
| Priority | When to action: Immediate, Near-term, or Can-wait |
Priority levels
| Priority | Meaning | Timeframe |
| Immediate | Large engagement gaps or high-impact zones | 1–2 weeks |
| Near-term | Moderate gaps or supporting improvements | 1–2 months |
| Can-wait | Minor refinements or polish | When time permits |
Detailed Zone Analysis (accordion)
Click the accordion to expand detailed analysis of each 5% scroll zone. This section includes:
• Executive Summary: Key findings with specific metrics
• Per-zone breakdown: Engagement metrics, reading behaviour, content elements
• Behavioural interpretation: What the patterns mean for user experience
Adding recommendations to Kanban
Each recommendation includes an Add to Kanban button. Clicking it:
6. Opens the project selector
7. Creates a task with all recommendation details pre-filled
8. Sets priority based on the recommendation’s priority level
9. Links the task to the page URL for easy reference
Once added, the button changes to show “Added” and becomes disabled to prevent duplicates.
Filter impact on analysis
Filters set before running the analysis affect which data the AI examines:
| Filter | Effect on analysis |
| Date Range | Longer ranges provide more data but may mask recent changes. 30 days balances volume with recency. |
| Device | Analysis focuses on selected device. Mobile and desktop often need different recommendations. |
| Referrer | Filters to traffic from specific sources. Useful for landing page optimisation by channel. |
Exporting the analysis
Click the export button (file icon) to download the analysis as a text file. The export includes all recommendations, zone analysis, and executive summary in plain text format for sharing or archiving.
Troubleshooting
Analysis fails or times out
Large pages or complex analyses may take longer. The system automatically retries up to twice if processing stalls. If analysis fails repeatedly, try a shorter date range or ensure the page is publicly accessible.
No data for selected filters
The page may have insufficient traffic for the selected combination of date range, device, and referrer. Expand filters or select a higher-traffic page.
Recommendations seem generic
The AI prioritises measurable, implementable changes. If engagement patterns are already good or gaps are small, recommendations focus on refinements rather than major changes.
Screenshot shows blank areas
Some page elements (lazy-loaded images, dynamic content) may not appear in screenshots. The AI notes when visual elements cannot be assessed and focuses on behavioural data instead.
FAQ
How long does analysis take?
Typically 30–60 seconds. Complex pages with extensive data may take up to 2 minutes. A progress indicator shows while processing.
What are the 5% scroll bands?
The screenshot is overlaid with markers every 5% of page depth (0%, 5%, 10%... 100%). This helps the AI correlate visual content with behavioural data at specific locations.
Can I run analysis on mobile and desktop separately?
Yes. Select the device filter before running analysis. Mobile and desktop layouts differ significantly, so separate analyses often yield different recommendations.
Can I ask follow-up questions about the analysis?
Yes. After analysis completes, a chat interface appears for asking questions about the results, requesting clarification, or exploring specific recommendations further.