Create Funnels

Introduction

Funnel is a Formula used to analyze the conversion rates between multiple steps in a flow. Kubit counts the unique users at each step, then divides it by the initial step to compute the conversion rates.

  • Each step can have one or more Events with Filters.
  • You can give a name to each step to make it easier to differentiate in the chart. If no name is provided, default step names such as the event name will be used.
  • You can add or remove steps using the controls on the left side.
  • You can also drag-n-drop steps to re-arrange their order.

Define a Funnel

Before we jump into it, a few words on the main Funnel controls at your disposal:

  • Steps: you can add up to 10 steps in a Funnel. You can define multiple events and filters for each step (up to 5)
  • Subject: provides the option of analyzing by User or Device (if applicable) as Subject
  • Partition:
    • Day: by default Partition the events by Day (all events have to happen within the same calendar day).
    • Session (OPTIONAL): for customers interested in Session analysis we can supply an additional option to tell Funnel that all steps have to happen within the same session.
    • None: this is useful when analyzing longer-term customer behavioral patterns. For example, any Subscription business would be interested in conversions from Trial to Subscription to Renewal and try to understand what makes users subscribe and then stay.
    • Conversion Window: all the funnel steps have to happen within the same time period defined by the Conversion Window. Supported Time Units for the window are Seconds/Minutes/Hours/Days/Weeks/Months/Years. Conversion Window can be used in combination with partitions Day or None so make sure not to use it in combination with Partition: Day whenever the Conversion Window length is greater than 1 day.
  • Attribution: a user may have more than one conversion path at slightly different times. This option can help fine tune how subjects are being counted in combination with Partition and especially when Conversion Window is being used.
    • Linear: when there is more than one conversion path, the subject will be counted on all steps of the longest path.
    • First: if there is more than one conversion path, the earliest event in each step is selected for the subject.
  • Group by (only applicable when there is a Breakdown ):
    • Starting: breakdown applied over the first step, subsequent steps do not have to fall within the same breakdown group.
    • Global: all the steps have to happen within the same breakdown group.
  • Order: there are four order options in Funnel. For three steps with events A, B and C:
    • Any: matches A, B and C in any order regardless of the event timestamps.
    • Sequence: matches A -> B -> C only when they happen sequentially, though there could be other events in between.
    • Exact: matches A -> B -> C in exact order without any other events in between.
    • Custom: allows you to mix Sequence and Exact steps in Funnel
  • Exclude Events: used to exclude events between steps when Order is Sequence or Custom in Funnel.

🚧

Edge case

When many events have exactly the same timestamp there could be indeterministic results in Funnel and Path. Same formula will returns different results because the event order is not guaranteed from one execution to another.

Uncover daily User habits

Let's say we have a Streaming application and we want to understand how many of the Users who Login proceed to generate Stream event and eventually give it a Vote Up during the same Day.

12041204

Explore long-term User experiences

Still analysing the behaviour of our Users in the Streaming application, let's have a look at how many of the Users who started a Trial in the last month proceeded to Subscribe. For the purpose we will use Partition: None, Order: Sequence group the results in the All time Time Unit.

12041204

Tips & Tricks

Re-order steps using drag-n-drop

Sometimes when you are working on a Funnel analysis and have started with order Any it doesn't matter how you define the steps, but when you realize you need to switch the order to Sequence it suddenly does. Here's a quick way to re-arrange the steps using drag-n-drop:

12001200

Compare conversion rate between multiple breakdown groups

If you want to compare the conversion rate between multiple groups along a period of time it becomes difficult to plot all the data in one chart, as we need to plot the time on one axis (1), the step conversion rates on the other (2) and then we need one more dimension for the Breakdown groups (3).

Thus, in Kubit we can balance those 3 dimensions and reduce them to 2 in order to provide insight from different angles like this:

  • only 1 breakdown group can be displayed for a time period (reduce groups)
15601560
  • multiple breakdown groups can be displayed for a single day (reduce time)
15601560
  • multiple breakdown groups for a period when only 2 steps of the Funnel are displayed (reduce steps)
15601560

Extending Time Window: extend the end date of the date range

Partition = None is used when analyzing longer-term customer behavioral patterns. For example, any Subscription business would be interested in conversions from Trial to Subscription to Renewal. This is where Extending Time Window comes in, as without it the conversion rates towards the end of the specified Funnel time range will become lower and lower due to the decreasing time window in which all the Funnel steps have to have happened and this could be misleading.

By setting Partition = None and adding an Extending Time Window you tell Funnel to group results by the date of the starting step event occurrences and look for the conversion steps until the end of the date range + the Extending Time Window.

27262726