What Is Funnel Analysis?
Funnel analysis is a method used to analyze the sequence of events leading up to a point of conversion. It lets product and marketing managers understand user behaviors and the obstacles encountered throughout the customer journey.
Not all prospects will become customers, and not all customers will immediately find the value of your product. Funnel analysis can help you pinpoint key events along the customer journey so you can conduct tests, improve the user experience, and increase conversions.
As an example, let’s say you’re trying to convert free trial prospects into paid subscribers. Your funnel might look like this:
- Step 1: Prospects open an email and discover an offer for a free trial.
- Step 2: They click on a CTA button to redeem the free trial.
- Step 3: Prospects create an account and use your product for free.
- Step 4: Prospects convert to paid customers after the free trial expires.
Many distractions or barriers can happen in between each of these steps, and there are likely patterns of behavior that can tip you off to what’s working and what’s not.
Although every business has unique goals, funnel analysis can be used to:
- Improve conversion
- Streamline the funnel
- Merge marketing and product teams
How to interpret and use data from Funnels
Let’s take the example above. We can visualize the data for that funnel like this:
- Emails Sent - 1000
- Emails Open - 300
- CTA button clicked - 150
- Free account created - 100
- Paid Customers converted - 20
Now this is a funnel and we can do certain analysis on it to understand the conversion on each step and see which step can and should be improved the most.
Conversion data in funnels might look like this:
Funnel analysis screen from Amplitude - a product analysis tool.
Seeing conversion data as funnels tells you a lot about which step has the biggest drop offs and you can then dive deeper into why so much drop off is happening on that step.
Example: Patreon Increases Subscriber Conversions with Funnel Analysis
Patreon provides creators, artists, and entrepreneurs with the opportunity to earn a living through donations. Users can “pledge” donations to creators on Patreon’s platform, and when creators win, Patreon wins. Patreon faced a conversion challenge—they needed to find new ways to incentivize monthly subscriptions to creator content.
Patreon discovered an opportunity to improve the pledge flow funnel through funnel analysis chart. Patreon tested a new feature called “blurred posts” to encourage more users to click through the pledge flow. These blurred posts concealed a portion of creator content, enticing users to delve deeper into the pledge flow funnel and ultimately subscribe. The result? Patreon was able to double pledge conversions on creator pages.
What is Cohort Analysis?
Cohort analysis is a type of behavioral analytics in which you take a group of users, and analyze their usage patterns based on their shared traits to better track and understand their actions. A cohort is simply a group of people with shared characteristics.
Cohort analysis is where you compare a specific cohort to another group of users. For example, let’s say you had a cohort of people who enabled push notifications during their first session. By comparing that cohort to another cohort, such as all active users, you can see whether that action affects how the notification-enabled users engage with the platform compared with everyone else.
The 2 most common types of cohorts are:
- Acquisition cohorts: Groups divided based on when they signed up for your product. Typically, the shared characteristics of this group of users offers an opportunity to measure retention and churn rates within a specific timeframe.
- Behavioral cohorts: Groups divided based on their behaviors and actions in your product. This type allows you to view your active users in different demographics and with different behavioral patterns.
Let’s look at some example to see how you might use cohort analysis.
Understanding New (and Underused) Feature Adoption
Meditation app Calm wanted to test their reminder feature. They noticed that a small set of highly engaged users actively used the feature, but the feature was buried in the settings menu.
Calm wanted to know if the reminder feature was helping increase engagement or if the users who were dedicated enough to wade into the settings were just already highly engaged, regardless of the reminders. The meditation company ran a test in which select users got a prompt to set a reminder after their first meditation session.
Using behavioral cohort analysis, the team could compare engagement among people exposed to the reminder prompt with relevant comparison groups rather than assuming existing reminder users were representative.
How to avoid misleading analysis
Define the funnel’s eligible population, event order, conversion window, and repeated-event behavior before calculating conversion. Segment by platform, acquisition source, geography, or lifecycle stage only when there is a decision you can take from the difference. A drop-off is a clue, not automatically a problem; some steps intentionally filter users.
For cohorts, keep the start event and observation window consistent. Compare D7 with D7, not a mature cohort’s 30-day retention with a new cohort that has only existed for a week. Remember that behavioral cohorts show association unless the exposure was randomized or causal assumptions are otherwise justified.
End every analysis with a decision: investigate instrumentation, conduct research at a step, run an experiment, or leave the experience unchanged.