Mastering Customer Retention Strategy
You've invested time and money into acquiring new users—but are you keeping them? With a deliberate focus on customer retention, you can improve retention at all stages of the user lifecycle to reduce churn and boost lifetime value.
Current user retention
Improving your current user retention is critical to creating a sustainable business. If your retention curve doesn't flatten out at some point, it will become impossible to sustain real growth.
At some point, even if you keep acquiring new users, poor retention will cause your overall growth to stagnate and even decrease.
Notice how the blue curve flattens off around day 7. Although there's some initial drop-off in the first seven days, a steady user base remains—these are your current users.
The goal of current user retention analysis is to move this baseline up to capture a higher percentage of retained users.
The green curve, however, never flattens off, meaning that your product is not attracting a steady base who continue using the product. This retention curve shape indicates you haven’t yet reached product-market fit.
Current users diagnostic
As defined, a current user was active in the previous period and is active in the current period you're measuring.
First, create your current user cohort, as covered in Chapter 3, and plot your baseline current user retention. Remember, you can use either N-Day (Return On) or unbounded (Return On or After) retention.
The example below shows a retention curve for a cohort of current users:
Investigate user properties and segment your retention curve
Once you create your current user cohort, look at user properties to get a high-level understanding of these users. Measuring the breakdown of key user properties can help you identify trends and groups of users you should study more closely.
You should also segment your retention curve by significant user properties (e.g., platform, location, attribution source) to identify differences to investigate. Refer to Chapter 4 for a refresher on segmenting by user properties.
Find behavioral personas of your current users
In Chapter 4, we introduced the concept of behavioral personas—each persona represents a distinct way of interacting with your product.
The purpose of identifying the personas of your current users is to understand:
- The value current users get from using your product.
- Whether there are distinct use cases.
- Behavioral patterns that might positively or negatively impact retention.
This section will discuss some examples of behavioral personas and principles for determining which personas to focus on.
Real-life example: Personas for a casual mobile game
One of our customers has a social casual game for mobile smartphones. The game matches players in real time and includes a social component with interplayer chatting.
When they analyzed their current users, they identified three core personas with high retention despite distinctly different behavioral patterns:
- High social only: Heavily use social features, but don't play many games
- High gameplay only: Mainly play games, but don't use social features
- High gameplay + high social: Actively play games and use social features
As you can see in the retention chart, the three personas have significantly higher retention than the baseline for all current users. In addition, the “high gameplay + high social” persona has the highest retention.
Although this data indicates that users who exhibit both behaviors have the highest retention, users who actively play games or use social features will still be retained at a much higher rate than the baseline. So even if the company focuses on increasing engagement with just one aspect of the product first (social or gameplay), they'll likely see significant retention gains.
Real-life example: Meditation App’s passive and core user personas
Another Amplitude customer has a mobile mindfulness app that provides meditation courses and scenes with calming background sounds.
Using the Personas feature, the app's product team identified three user personas:
- Listeners: Primarily listen to and swipe through different scenes
- Meditators: Completed more meditations than average
- Alert Savers: Activated a feature that sends a Daily Reminder to meditate. These users also completed several meditations per week on average.
The Alert Savers persona was particularly interesting: Only ~1% of users were setting an alert. This feature was buried deep within the app's Settings screen, so very few users discovered it—but these users had very high retention compared to other groups.
Using the power user, core user, and passive user framework, the company classified:
- Listeners = Passive users
- Meditators = Core users
- Alert Savers = Power users (because they used a “power feature”)
Comparing retention curves of different personas
This company also compared the retention curves of “Listeners” and “Meditators.” Both personas had similarly high Day-N retention 30 days after the current period, although Listeners had slightly lower retention.
To see how these personas might differ in retention longer-term, they looked at weekly retention for the next 24 weeks to help uncover some larger differences: Alert Savers have the highest long-term retention, followed by Meditators, then Listeners.
Based on these retention graphs, Listeners are a fairly active Passive persona, but still have lower retention than Meditators long-term. In addition, Alert Savers who set a daily reminder to meditate have the highest long-term retention at 24 weeks. To increase overall retention, the company should try to convert Listeners into Meditators and encourage Meditators to set a daily reminder and become Alert Savers.
Dig deeper into your personas: Product Analysis Toolkit
Once you identify your current user personas, you can use some or all of the analyses in the Current User Worksheet to better understand how these users behave. This will help you identify improvement opportunities and additional drivers of current user retention.
Reference the Product Analysis Toolkit in the Appendix if you need to review these methods.
Critical event stickiness
It's also essential to look at critical event’s stickiness, or the frequency (number of days) a user is active or did a specific event.
- Amplitude Customer Example: App that helps users find and book nearby fitness classes.
- Stickiness Graph 1: Measures each day user performed critical event—when a user books a class— within a month
- Stickiness Graph 2: Measures each day the user performed any activity—like browsing classes or checking a class schedule—within a month
Stickiness Graph 1 compared to Graph 2 told our Amplitude customer that although a high percentage of each of the three personas are opening and engaging with the app 15 days a month, the percentage booking appointments frequently is much lower.
Stickiness Graph 1 also helped this Amplitude customer uncover that Persona 2 and 3 are booking appointments at a higher frequency than Persona 1, and that it’s more valuable to focus on increasing the number of users and improving the experience for Personas 2 and 3.
Discovering the drivers of habit formation
Once you successfully onboard and demonstrate value to new users, they’re returning regularly. This is habit formation.
Studying current user retention is about understanding the factors that encourage people to form a habit. By studying current users, you'll pinpoint indicators of habit formation. You can then apply this to encourage more new or resurrected users to form habits. To help make this a repeatable process, we will show you how to look for behavioral drivers that tip the scale for habit formation.
To understand what gets new users to become current users, you dig into the behaviors that drive that transition.
Identifying the behavioral drivers that move users through phases is similar to identifying your “a-ha” moment. Traditionally, the “a-ha” moment is an action a user takes early in their experience that makes them much more likely to retain. The most famous example is when Facebook identified that users who added at least seven friends in their first ten days are more likely retained.
Interested in other companies' “Aha” moments? Amplitude customers speak to the data discovery that transformed their role and company in this eBook.
However, you can apply this concept of important behaviors to any stage of the user life cycle, not just for the “aha” moment of new users. To identify the drivers of habit formation, find an action or set of actions that separates users who complete Habit Formation from those who don't. For action(s) to qualify as a driver:
- Most users who complete the action(s) form a habit and become current users
- Most users who do not complete the action(s) churn before becoming current users
How to find your drivers of Habit Formation
In this section, we'll cover how to identify the behaviors that drive users to complete the habit formation phase.
Step 1: Create a base cohort of users who were retained during the Habit Formation period.
The table and images below show the Habit Formation period that you should analyze based on the usage interval you calculated in Chapter 2.
Step 2: Create a retained cohort of users who were retained in the next interval after the Habit Formation phase. These are current users who successfully formed a habit.
Step 3: Create a dormant cohort of users who were in the base cohort and were not retained in the following period.
Step 4: Compare your retained and dormant cohorts to find behaviors in the retained cohort but not in the dormant cohort.
You can do this by:
- Brainstorming actions you think might be essential drivers and measuring the percentage of users in your retained and dormant cohorts who performed those actions.
- Talking to users from both groups to get qualitative data.
- Watching user replays or looking at user activity sequences from both cohorts.
For example, a music streaming product could hypothesize that important actions might include: playing songs, creating playlists, favoriting songs, and so on. The team would then evaluate any differences between the retained and dormant cohorts in the frequency they perform these actions.
Step 5: Once you've formed some hypotheses of potential drivers, measure the difference in retention between users who perform that action and those who don't. This will help confirm whether or not performing that action correlates with higher retention.
Below you can see that users who favorite at least one song have significantly higher retention than users who do not.
Apply what you've learned: Get more users to form habits and become current users
Once you've discovered your drivers of habit formation, you’ll know the milestones that new users need to get through to increase the odds that they’ll continue using your product as current users.
To implement these insights, consider getting more users to pass these milestones during their early experience. For the gaming company we previously discussed, these could include:
- Sending push notifications when a user's social connection is active in the game, encouraging them to join them.
- Presenting users with a reward, like a badge or in-app currency, once they've used the social feature a specific number of times.
By experimenting with a few methods, you can uncover the most effective ways to get users across the habit formation threshold.
Discover drivers from passive → core → power personas
As we discussed in Chapter 4, you can classify personas as passive, core, or power users. Earlier in this chapter, you identified more active and valuable personas than others.
Remember that for the mindfulness app, they found a passive persona of Listeners and a core persona of Meditators. To increase core usage of their app, they should try to get more Listeners to become Meditators.
So, how do you get a passive user to become a core user or a core user to become a power user? Just like we identified drivers of Habit Formation, you can identify behaviors that drive people to become core or power users. To do so, use the same process as before.
Now that you’ve completed the current user retention analysis, summarize what you’ve found and form some hypotheses to test.
Here are some key questions to ask yourself:
- What are the key action(s) you identified as drivers of Habit Formation? What methods can you test to get more new users to cross those thresholds?
- Who are your passive, core, and power users? How are they different? How can you convert core users to power users?
- Did your behavioral persona analysis reveal any use cases you didn't expect or didn't think were very important? How might you improve or tailor the experience for those users?
- Are some of your personas more important to your main business objective, like revenue?
- How can you get more users to convert into one of your core or power user personas? The most significant improvements can come from targeting users who are not well-retained and getting them to perform the same actions as your power behavioral personas.
Track improvement over time
As you test some of your hypotheses and pilot new ways to improve your current user retention, it's essential to keep track of metrics to see what is and isn't working.
Keep the goals of current user retention in mind as you form your metrics:
- Get new users to form habits and become current users.
- Get current users to become core users and core users to become power users.
We suggest tracking these metrics over time to measure your progress:
- The size (in absolute numbers) and percentage of your total active users that consists of your current users (as calculated via Lifecycle or manual analysis).
- Retention over time of all current users and of each behavioral persona.
- Size and percentage breakdown of your important behavioral personas. Are you getting more people into important personas?
- Stickiness over time for critical events, showing you any changes in how current users behave in the product.
- Conversion rate over time through your critical path funnel.