Use the Analysis overview
Open Analysis from the main navigation and select the project you want to inspect. The overview is project-wide and is not limited to visitors enrolled in an experiment.
Dashboard cards
The default dashboard contains the following cards:
| Card | What it shows |
|---|---|
| Visitor summary | Unique visitors, sessions, and average sessions per visitor. |
| Visitors by device | Desktop, tablet, and mobile visitors as percentages and counts. |
| Top 5 countries | The five leading known countries by visitor count, shown in a ranking and on a map. |
| New vs Returning | The split between visitors seen in one session and visitors seen in more than one session. |
| Page journeys | Session paths through the website as an interactive sunburst diagram. |
| Top traffic sources | The five leading referring sources, with their share and visitor count. |
| Hypotheses | Ranked CRO opportunities generated from the collected project data. |
Explore page journeys
The center of the Page journeys chart represents the start of a session. Each ring moving outward represents the next page in the journey. Segment size is based on the number of sessions following that path.
- Hover over a segment to see its page label and session count.
- Select a segment to focus on that branch and update the legend to its next steps.
- Use the expand control in the card header to open the chart in a larger dialog.
Arrange the dashboard
Select Manage cards to customize the overview:
- Drag cards into the preferred order.
- Use the eye control to show or hide a card.
- Use Toggle All to show or hide every card.
- Use Reset default to restore the original order and visibility.
You can also drag visible cards directly on the dashboard or hide one from its card menu. The arrangement is saved in the current browser and is reused when you switch projects.
Data maturity
Select Data maturity above the dashboard to see how much evidence is available for analysis. The phase is based on collected sessions:
| Phase | Best suited for |
|---|---|
| 1 — Bug detection | Finding obvious breakage and strong friction signals. |
| 2 — Pattern detection | Identifying repeated behavior patterns. |
| 3 — Full analysis | Drawing broader CRO conclusions. |
| 4 — Small-effect tests | Evaluating subtler opportunities and smaller expected effects. |
Data maturity communicates how confidently the available evidence can be interpreted. It does not prevent you from opening the dashboard when fewer sessions are available.
Generate CRO hypotheses
The Hypotheses card turns the collected analysis into ranked, actionable ideas:
- Optionally select a detected page group to focus the report. Leave it unset to analyze all pages and automatically select the leading issues.
- Select Generate CRO report.
- Select a hypothesis to open its full details.
Hypotheses are ranked using impact, confidence, and ease. Each item includes a confidence tier, a falsifiable hypothesis, a proposed change, target KPI, supporting evidence, and known blind spots. Confidence labels include Confirmed bug, Likely issue, Validated, and Collect more data.
The card records how many sessions the report used and when it was generated. It also shows how many sessions have arrived since then. Select Refresh report to regenerate it from newer data; the page displays a reminder when the dataset has grown enough to consider refreshing.
Ask questions about the analysis
Use Ask about this at the bottom of the page to ask questions about the selected project's CRO data and hypotheses.
- Press Enter to send a question or Shift+Enter to add a new line.
- Minimize the chat, pin it beside the dashboard, or open it full screen.
- Conversations are kept separately for each project.
- A new conversation uses the current CRO report as its starting snapshot. If the report is refreshed later, the chat shows that it refers to an earlier report and offers Start new chat.
Treat generated hypotheses as prioritized investigation and test ideas, not guaranteed outcomes. Review their evidence and blind spots, then validate promising changes with an experiment.