About the Sankey View
The Sankey diagram is a flow map that shows the paths users actually took through a Task, not just how often each action was performed.
Think of it like a river that splits into streams. The wider the stream, the more users flowed that way. Where streams narrow or stop, users dropped off or took a detour. This makes it easy to see at a glance which journey is most common, where users struggle, and where they give up without reading a single table or report.
- Use the Action chart for action-level event counts ("How many times 4th action was performed")
- Use the Sankey for flow, sequence, and drop-off analysis
How to Read the Sankey
- Flows (connecting lines): Flows show transitions between actions. The thicker the band, the more users took that path.
- Nodes (bars): Each node represents an action defined in your Task. The height of a node is proportional to the number of events recorded for that action. Event counts are shown on hover for each node and flow.
- Columns: Each column represents a step in the user journey, how many actions deep users were at that point.
For example, "Path X took users only 3 clicks to complete the task"
- Rows: Represent actions took at that each step.
For example, "Most common 3rd actions for users were X, Y and Z"
By default, the diagram shows the 3 most common paths. Less common paths are grouped under "Other events." Increase the number of rows to reveal more detail.
Drop-off nodes: Grey nodes indicate where users stopped and did not continue the Task.
For example, drop off in the 3rd column means that we stopped receiving Task events after second user action.
How to Use It
- Find the dominant path: Follow the thickest bands from left to right. This is the most common journey users take through your Task.
- Spot drop-offs: Look for large grey drop-off nodes. The column where they appear indicates how far into the Task users got before stopping.
- Identify friction by looking for:
- A flow splitting into many branches at one step (confusion or unclear UI)
- Users returning to the same action across multiple columns (rework or missing context)
- A large share of users reaching "Other events" (increase rows to investigate)
These patterns could indicate confusion, missing guidance, or unnecessary effort.
- Find automation potential by checking for:
- Thick, repeated paths - actions many users do manually, again and again
- Loops - the same steps repeating across columns
- Long paths for simple outcomes - too many steps to reach a goal
These patterns suggest steps that could be automated or skipped entirely.
- Compare different paths: Compare paths by thickness (how many users follow them), length (how many steps they take), and drop-off rates along the way. This helps you answer: are users taking different routes to reach the same outcome, and which one is more efficient or less error-prone?
- Identify the best path: The best path is usually the top (fastest) path, the thickest path, and the one with the fewest steps and lowest drop-off. This is the journey to protect, promote, and optimize.
Multiple starts
Task can be defined with multiple starting points in the Editor. If your Task has multiple starting points, the Sankey shows which one users triggered most often, helping you understand where users actually begin the workflow.
Different Task versions
If your Task has been edited structurally (actions added, removed, or reordered), you will see different versions of the Task in the analytics. We've added filtering by Task version to avoid mixing data from different versions of the same Task.
By default you will see the latest version and data from when it this latest Task definition was active.
Good to Know
- Data refreshes hourly.
- The Sankey reflects behavioral data, not your Task definition. Users may take actions in a different order than you defined them.
- Numbers in the Sankey may not exactly match the Action chart. This is expected. The Sankey applies readability filters: consecutive repeated actions are collapsed, actions can repeat in a loop, and very unusually long sessions (top longest 10%) are excluded from the visualisation. Therefore, use the Action chart for precise event totals (e.g., "total amount of clicks and users on 6th action in the selected period").
