Paths enable you to assess how combinations of attributes affect the likelihood of your predicted goal.
Paths are represented in a decision tree visualization. The first level (left most column) can be interpreted similar to the Benchmarks feature - it assesses the threshold in user activity for a specific attribute that lead to the highest increase in probability of your predicted goal.
In the screenshot above, you'll see that users who performed the action 'modelstab-nullview' last week > 0 times had the higher likelihood to perform the goal.
The second level of the Paths diagram splits your users into two groups - between those who performed the attribute in the first level above and below the respective benchmark. For each group, Paths then evaluates the next attribute that has the highest likelihood of increasing the probability of your goal.
In the example above, we accordingly see that users who performed the action 'modelstab-nullview' last week > 0 times AND 'allmodelsactivity' > 1 times had the highest probability to perform the goal.
You can follow paths for up to 4 levels, representing 8 distinct combinations of attributes.