Menu

Path Analysis

1. Overview

Path analysis is used to analyze the proportion of events at each step in a sequence of user behavior events, starting from a specific event, or to analyze the proportion of events leading up to a specific ending event in the preceding steps.

Through path analysis, you can understand the distribution of user behaviors in a sequence of continuous events.


Path analysis can address the following analytical needs:

  • Through which paths do users achieve payment conversion?  
  • Why do users deviate from the expected path? What are the reasons?  
  • What are the differences in behavior paths among users with different characteristics? How are these differences generated?  
  • Which path is the most frequently visited by users? At which step are users most likely to drop off?  


2. Usage Guide

The main configuration interface for path analysis conditions is as follows.

2.1 Start or End Events

The start or end events are the basis for selecting the analysis method. You can choose to analyze events starting from a specific event or ending with a specific event.

  • When analyzing events starting from a specific event, the results will be calculated with the specified event as the starting step, followed by a sequential listing of the events that occurred in subsequent steps.  
  • When analyzing events ending with a specific event, the results will be calculated with the specified event as the ending step, followed by a listing of the events that occurred in the steps leading up to the ending event.  


2.2 Events Involved in Analysis

Participating Events are events selected to be included in the path analysis, apart from the specified start and end events. During the calculation, all selected participating events, along with the start and end events, will be used as the participating events for session segmentation. In the final path analysis results, only the distribution of participating events across each step will be displayed.

  • Participating events support batch selection, as well as batch modification and cancellation.
  • A maximum of 20 participating events can be selected.


2.3 Filter

Click "Add Single Event Filtering" below the participating events to add individual filter conditions for a specific event among the participating events. Only events that meet the specified filter conditions will be included in the path calculation.

Click "Add Condition" to add unified filter conditions for all events. These filter conditions will apply to all participating events, and only events that meet the conditions will be included in the path calculation.

2.4 Session Interval

The maximum time interval between two adjacent participating events is called the session interval.

If two adjacent events occur within the interval time, they are considered part of the same session. If the time interval between two events exceeds the set interval time, the session will be split. The first session will end, and a new session will begin when a start or end event occurs again.

For example, let's assume the user performed the following behavior events:

EventA1A2FBDA3A4CEF
Time11:0011:0511:0811:1011:3512:0012:4013:0013:1013:30
Time Interval5525254020

Set the conditions as follows: Start with Event A, Participating Events: B, C, D.

  • With a 60-minute interval, since the time intervals between all participating events do not exceed 60 minutes, all events are grouped into one session: A1-C.
  • With a 30-minute interval, since the interval between A3 and A4 is 40 minutes, which exceeds the set interval time, the original single session will be split into two sessions: A1-A3 and A4-C.

Therefore, the session interval affects how sessions are segmented, and it can be adjusted appropriately based on analysis needs.

2.5 Chart Display

The results of path analysis are presented in the form of a Sankey diagram, which visually displays the proportions of node events at each step.

When selecting the analysis method that starts with a specific event, the leftmost step represents the starting step.  

When selecting the analysis method that ends with a specific event, the leftmost step represents the ending step.

You can flexibly choose the number of steps and nodes to display on the chart based on business needs.

Steps and Nodes

  • Steps: The number of steps displayed to the right, starting from the beginning or ending step.  
  • Nodes: The number of events independently displayed within each step. Only nodes with a high proportion will be displayed individually, while nodes with a low proportion will be grouped under "Others." The node count for each step does not include "Others," "Churn," or "New."

Churn or New Nodes

  • Churn: When analyzing events starting from a specific event, users who do not proceed to the next step are considered dropped off and are summarized under "Churn"  
  • New: When analyzing events ending with a specific event, users who start participating in events from a specific step are considered new users for that step and are summarized under "New." Note that "New" here does not refer to actual new user behavior in the product or application but rather to users appearing for the first time in the current behavior path analysis. This distinction should be noted.

3. Permission Explanation

Operation Permissions: Super Admin, Group Admin, Analyst, Regular Sub-account with assigned permissions.

Entry: SolarEngine - Product - Analysis - Path Analysis

Previous
Distribution Analysis
Next
Users Analysis
Last modified: 2025-05-29Powered by