Retention Metrics

“Usually you can see a nice correlation. If usage drops the willingness to pay goes down and people are likely to churn.” (Markus Lang)

Our next metric that we will cover is retention metrics. Retention metrics is the one that gives an indication of the quality of growth.

Customer retention rate

Theory

Indicated as percentage, this metric shows how many users retain after their initial engagement with the product (Datarockets, 2019). A useful concept to keep in mind are cohorts. Cohorts are simply groups of customers who, for instance, signed up for your product in the same month (Skok, D., 2019). When looking at companies with network effects - meaning the more user the higher the value — the newer cohorts which joined recently and are benefitting from a large network should show a better retention rate than older cohorts. However, what can be seen in practice is that these older cohorts often show a better retention rate because they are early adopters and hence, highly motivated users and thus stick around longer (Jin, L., Coolican, D., 2018).

Application

For the customer retention rate (CRR), subtract from the number of customers you have at the end of the given time period (E), the number of newly acquired customers (N). Divide this number by the number of customers you had at the start of the period (S) (Datarockets, 2019).

Practice

Retention rates can be measured in terms of activity levels or also payment retention cohorts.

Active Users

Theory

You can look at your daily active users (DAU), weekly active users (WAU) or monthly active users (MAU). In order to find your “best” users, you can segment them based on demographics or behavioral data. A good way to understand user behavior is to further dive into the behavioral segmentation and look at the so-called power users. It shows you particular core actions that were performed in any given time frame. Especially businesses with network effects should see an increase of core actions performed as the company matures and more cohorts join. This means that users get a higher utility due to a higher overall number of users (Jin, L., Coolican, D., 2018).

Depending on the product it makes sense to track different types of active users. Two most common ones are:

• Daily Active Users (DAU): This metric shows you how many unique users engage with your product on a daily basis (Datarockets, 2019). However, it does not reflect on the quality of the users nor on their retention. A spike in DAU could happen for instance only due to a customer acquisition campaign and not because suddenly people start to love your product. Thus it is important to measure core daily activities. Calculating Core Daily Actives is easy: simply look at the users statistics on a given day and calculate how many of those have used your product in the weeks before (Egan, J.).
• Monthly Active Users (MAU): This metric shows how many unique users engage with your product on a monthly basis (Datarockets, 2019). There are different ways to interpret MAU.

There are 4 splits to identify where your user growth comes from: Signups, Resurrection, Existing User Churn, New User Churn, (Egan, J.). When splitting up the user growth in these four categories it is important to track the total number of users over time to identify outbreaks and unusual activities.

Application

• Pay special attention to the definition of active users which may vary from business to business. For instance, you can choose to define active users as users who open the app once per day/week/month or as users who made a purchase conversion, for instance (Datarockets, 2019).
• Strong activation rates are necessary for scaling the service and to achieve sustainable growth. That being said, it is important to split the different segments in order to understand the characteristics of the best and worst performing segments (Egan, J.).

Practice

MAU is often used for marketplaces and subscription-based businesses whereas DAU is often used for ad-based businesses since their daily usage is heavily influencing their revenue stream (Datarockets, 2019). A good benchmark for DAU/ MAU is > 50% since this signals that the product is part of a daily habit (Chen, A.).

Churn Rate

Theory

The churn rate indicates the percentage of active users who stopped using your product (Datarockets, 2019). Over time, you want to minimize this number as much as possible. The solution for this is to achieve a negative churn which happens when the expansion revenue from existing customers exceeds the lost revenue from churning customers (Skok, D., 2019).

You can do this by implementing a variable pricing scheme that lets the user expand the usage of the product and hence pay you more. A second option would be to up or cross-sell them to other products. Look at the two graphs below which illustrate the principle of negative churn nicely. On the left side, you see the case of a monthly moderate churn rate whereas on the right side you see what happens when a few people churn but you manage to substantially increase the revenue from your existing customers (Skok, D., 2019). Finally, it is essential to find out why customers churn in order to tackle the issue which causes them to leave (Law, 2016).

Application

• Start by analyzing your overall and your churn for the individual customer segments for both customer and MRR churn (Skok, D., 2019).
• A useful tool for analyzing churn is cohort analysis. A cohort is simply a fancy term for a group of customers who signed up for your product in the same month, for instance. This helps you to see how early customers churn or if the churn rate stabilizes after a while. In order to minimize churn, you can make improvements to your product.
• When conducting an in-depth analysis of your churn rate look into revenue AND customer churn! If you have a churn rate of 10% it could be that 9 small and 1 big client left but you can also have the same churn rate of 10% with 9 big clients and one small one leaving you. This would have a completely different impact. To sum it up, monitor the percentage of the retained cohorts over time as well as the percentage of MRR retained over time (Skok, D., 2019).
• Reasons which cause a high churn rate could be that you are not meeting your customers’ expectations, your product does not offer enough value for the user to stick around and pay, your customers did not fully adopt the product and are not using key sticky features or the problem of who you are selling to (Skok, D., 2019).
• If churn is your problem there is only one way forward: go out in the field and talk to your customers to understand the underlying problem. If you are unable to fix churn you are solely filling a leaking bucket (Skok, D., 2019).

Practice

Churn rate in particular also needs to be tracked if you are selling to enterprise businesses. If a large number of customers who have been with you for the last 2–3 years cancel their contracts, that’s a bad sign.

Cohort Activity Heatmap

Theory

The cohort activity heatmap is building upon a cohort analysis. As already stated in the name of the metric it shows the activity levels of your cohorts (Egan, J.). It differs from the usual cohort analysis in a way that shows real-time behavior instead of reactive data on your customer churn. This means that you are able to see when your user engagement drops before they churn and you can take immediate actions.

Application

In order to build a cohort analysis you will need to include the following variables into your matrix:

• X-axis: number of days passed and new cohorts
• Columns: cohort joined on that day
• Width of column: number of users in cohort
• Y-axis: number of days passed and cohort activity level
• Colour: Activity level of cohort

In order to determine the activity level of your cohort you can implement a customer engagement score and score users who perform certain key actions higher than others. For instance, people who are posting a photo on Facebook are more engaged and less likely to churn than people who log in and view one page. Allocate more points to the features you think are the “stickiest” (Skok, D., 2019).

Finally, it is all about identifying cohorts which stick for a long period of time and are frequent users. Test different campaigns and product alterations to find out which activities attract highly valuable cohorts.

Practice

Ideally the cohort flattens and >20% of the cohort stick with your product for a long period of time.

Tracking user retention is extremely important since users who come back on a, for instance, daily basis are more likely to renew their subscription. Usually, there is a good correlation if people stop using the product the majority also stops paying for it.

Across seed and series A cohorts need to be monitored closely. It can be distinguished between user, order and revenue cohorts. Here, it is easier to monitor the percentage changes versus month zero. Only for revenue it is interesting to see absolute numbers. Another interesting thing to look out for are power user cohorts.

Already at the pre-seed stage it is essential to track retention, cohorts, DAU/ WAU/ MAU, and how the users engage with the app.

What is important to say is that the use of the activity heatmap is not suitable for all startups, though. Still there are ways to set up cohort analysis which generate maximum insights.

What do you think of these metrics? Have you ever had a chance to apply them? We can’t wait to hear your thoughts!

Get the full publication on startup metrics here!