Funnel analysis has become the cornerstone of event-based analytics. A funnel analysis is a predetermined sequence of events, through which conversion rates are tracked at each step. Using Amplitude's Pathfinder and behavioral cohorts features, you can easily identify and set up insightful funnels. This article teaches you how to set up a funnel, analyze data, and take action on the results to improve your funnel conversion rates.
Identify funnel events: Which ones should you include?
Funnels should track the flow of users along a critical path in your product. Each funnel step is a distinct action a user can take within a flow you want to analyze.
For example, if you were measuring an onboarding funnel, you may track the following events:
For an e-commerce platform, tracking a purchase funnel might involve the following sequence of events:
Add to Cart→
These examples work great if you already know the paths your users take. However, it's impossible to know every possible navigation route within your platform, especially since the ones your customers are using may at first seem counterintuitive.
The Pathfinder feature shows the most commonly-taken event paths in your product, either following a specific start action or preceding a specific end action, and helps you discover the alternate navigation routes your users are already taking.
Imagine your product is a music streaming app with a social component. After using Compass, you find that
AddFriend is the event most correlated with retention, your most important KPI. A key focus area might be to identify how you can get users to add more friends. The next step will be to understand the different paths users take when adding friends, so you can encourage that behavior.
Let's start by trying to understand the actions users are taking immediately before triggering the
AddFriend event, by analyzing this ending-with flow:
The graph shows that
PlaySong is the most commonly-triggered event immediately before the
AddFriend event. In this example, assume that prior to creating this graph, you made a list of the events we'd expect to show up as the top events in this ending-with flow. You weren't expecting
AddToList here, which means you should investigate further.
Let's look at this in the other direction. What events are users triggering immediately after adding a friend? To answer this question, build a starting-with flow:
It is intriguing that
AddToList seem to be happening so close to each other, since they're all parts of different overall path flows for your app. Would more users trigger
AddFriend if you could get them to play or favorite more songs?
Design the funnel
Exactly how important is the exact sequence of events in your funnel? Is it enough that a user triggered all the events within a specified length of time? Or must they be done in a particular order?
When setting up your funnel, you will have to decide whether to use Any Order or This Order mode. The former counts as converted all users who trigger the funnel steps in any order during the conversion window, which you can set anywhere between one and 90 days. The latter setting requires users to follow the exact sequence of events in order to be counted as converted. The conversion window is potentially much shorter for this mode: you can set it to as little as one second.
Read more in our Help Center article about building a funnel analysis in Amplitude.
Imagine you've been sending push notifications to certain users to get them to play songs. However, now that you know that a lot of users trigger
AddToList close to each other, it's worth understanding whether sending those push notifications actually makes those users more likely to also add more friends. If so, that would be a great insight, and would validate the hypothesis that push notifications have a complementary effect on other major KPIs.
Read more about the semantics of our funnel conversion window here.
Build the funnel
For this first funnel, make an "Any Order" Funnel, since you are only concerned with whether the user took all the required actions, and not the sequence in which the user took them. Also, you still don't know if more people trigger
AddToList first; you'll want to capture both sets of users to determine whether there is a cluster of users who trigger all four of these events together.
In other words, this first funnel is just to get a sense of whether this hypothesis merits further examination.
Set the conversion window to one day, as that should be an appropriate length of time for someone to take the following four actions:
Receive Push Notifications →
Play Song →
Add To List →
Add a Friend
Once set up, your funnel should look like this:
And here are the funnel results:
This funnel has a 78.6% conversion rate, which means that 78.8% of all users who entered the funnel took all four actions within one day. This is a strong conversion rate and looks interesting enough to further analyze, but it does not deliver any actionable information on its own.
Next, consider comparing two different of users for this funnel. You can compare users who got a push notification and then played a song to those who didn't play a song after receiving the notification. The goal is to see if you can get users to increase their network of friends more quickly by convincing them to play more songs via push notifications.
To get that list of users, build a second funnel with the following setup:
This funnel is intended to generate a list of users who received a push notification and then played a song within five minutes of receiving it. Since the order of events matter in this case, compute this funnel in "This Order" Mode.
The image below shows the funnel results:
To create the two cohorts described previously (users who played a song and users who did not play a song after receiving a push notification), use Microscope to create a cohort of users who converted from this funnel. Then name the cohort “Played song from Push”.
Now, revert back to the initial funnel and segment the funnel to compare the two different cohorts of users: Those in the “PlaySong from push” cohort (blue segment) and those not in the cohort (green segment).
The results here are telling.
75.5% of the users who played a song within five minutes of receiving a push notification also triggered both
AddFriend within one day. This is a 119% increase over users who did not play songs from push notifications.
In other words, a user is more than twice as likely to trigger
AddFriend if they a) receive a push notification, and b) play at least one song after receiving the notification.
Next, you might want to see if there are particular trends on conversion over time and look for critical insights in the property distribution of the
PlaySong event. From the
SongSource event property for
PlaySong, it seems that the contribution of
Radio is very similar.
Trends in conversion over time seem to stay consistent for both cohorts (the graph below is following the same color pattern as above, so blue color segment represents users in our cohort and green represents those not in the cohort).
Overall, we know that users who receive push notifications and play a song within five minutes are three times more likely to add friends, but there are no critical insights from property source distribution (for the type of song they play) or conversion over time.
Acting on the funnel insights
Your next goal should be to create a retention loop where you can get more users to come back and take the desired action—add more friends—which is correlated with retention, your main KPI in this scenario.
Go back to your funnel and use the Microscope feature to make cohorts of users who dropped off from the funnel at critical points, like
AddFriend, and message them to take the corresponding actions.
You can use an in-house messaging platform or one of Amplitude's push notification partners like Urban Airship or Kahuna to message these cohorts.
We hope this article helps you understand how to get the most out of funnel analysis on Amplitude. Funnels can provide some incredible insights if you use them in conjunction with other features like Pathfinder, behavioral cohorts, and the Compass chart.