Last week, one of the founders I have been fortunate to work with for a couple of years posted this chart with the following comment: ‘One of the most fascinating facts about Flo is that we get more than half of the revenue from users who installed the product more than a year ago. Close to 25% of the revenues come from users who have used the product for three years or more. And the share of such older cohorts is growing. This is the power of products with good retention, and you may rarely see such a picture among B2C apps.’
This chart with cohorts articulates well the power of subscription models when coupled with a best-in-class product that people use continuously over time. The retention of users and the recurring nature of the business model not only make it a highly profitable model, but also compound over time. If you have this in a very large market, which is the case here, the sky is really the limit. Not only can you add many more users over time, but also offer them more features to improve their user experience and value extraction.
Cohorts unveil the ground truth behind any business.When people join our team, I usually tell them that cohorted usage and purchase data present the ground truth in any business, regardless of the business model. It is one of my favorite measurements in assessing an investment opportunity, and the data point we spent the most time on. As I had been working on such an analysis with the founder of one of our younger companies last week, I thought I’d share some insights with the broader OE community...
While relatively straightforward to build at first sight, it is never as easy as anticipated. First, you need to find the data to build the cohorts. In the early days, it is a relatively manual and, to some extent, creative process. Every cohort gives you new insights that push you to search for more data...
What is a cohort analysis?A cohort analysis isolates a group of people who share a common characteristic or experience within a defined period. They demonstrate how much value these customers can produce over time and inform a business’s willingness to pay to acquire a customer. The sum of all cohorts gives you the aggregate numbers of the business. In other words, financial models that make predictions about future revenues are effectively ‘roll ups’ of new and existing cohorts of customers.
It is a powerful method used to analyze groups of customers and their behavior over time. This can be done with a variety of metrics. Grouping is usually done based on when a customer has signed up or converted into a paying customer. Depending on the size of the data at our disposal, other additional groupings are by geography, acquisition channel or sales team, contract duration, or customer segment. By conducting a cohort analysis, you can track customer-specific behaviors and usage, retention, churn, and revenue over time.
The chart above shows the example of a fictive young company that has 7 months of data on paying customers (here is the underlying xls sheet with various examples of a consumer subscription app). It shows you the revenues generated (x-axis) by clients acquired in a specific month (y-axis) for each month following their conversion into a paying customer. As you can see, I usually suggest adding the marketing budget spent to acquire those customers to understand better the evolution of the acquisition dynamics.
Another way to look at these cohorts is to focus on the retention of those revenues month over month. It’s often more interesting to look at percentages (rather than the absolute number of customers or revenues) as it is easier to compare retention rates across different cohorts with different numbers of users and revenue levels.
Here is the graphical representation of such a cohort analysis.
Another nice way to visualise cohorts is to put them on top of each other. Here is an example of revenues generated by a specific monthly cohort.
Cohorts can be presented in many different ways. The key is actionable outcomes.In the previous example, we focused on the revenue evolution. Still, the same exercise should be done with various metrics to have a complete picture: usage, number of customers, etc. The chart shared at the beginning of the post has a slightly different, more dynamic approach by focusing on the revenues generated each month by a particular customer group that evolves over time based on their onboarding date. Such an analysis requires you to have enough history to set it up. It is not helpful for younger companies.
It is essential to compare apples with apples. In the case of subscription businesses for example, keeping annual and monthly plans separate is essential. Mixing customers with different contract durations leads to highly inaccurate results. These customers on annual plans cannot leave during their first year, so they artificially increase your retention rate. Therefore, segment your customers into different plans with separate cohort analyses.
The most important is to find actionable outcomes. Some examples:
Cohort analysis helps you identify when and where you’re losing customers. Why are customers churning? Is there a specific time-limited reason, or is churn consistently high? If the latter, it indicates that you probably don’t have PMF yet.
Cohort analysis is a great way to forecast customer lifetime and estimate CAC payback times. This will help you set up your acquisition strategy.
If you have distinct customer segments, you might discover that one has better metrics. You should focus on that segment in the coming months.
Cohorted retention graphs effectively show which new product features, software releases, or acquisition campaigns brought better or worse customers.
What makes cohorts so useful?Simply looking at topline revenue growth or a usage metric of a business over time doesn’t tell you much about your ability to retain users and turn them into happy long-term customers. If you’re acquiring a lot of new customers quickly, metrics like ‘usage from repeat users’ or ‘payments from repeat customers’ may go up nicely, even if your retention rate sucks. The overall picture of the ‘machine’ might look nice, but not the underlying evolution of your customers. This is one of the reasons why investors are so obsessed with churn.
This is why it’s essential to use cohort analysis, which is all about retaining engagement, customers, and revenues. By understanding how various customer groups stick around (or don’t), you can get a much better idea of what’s really happening with retention. The evolution of cohorts can also help you determine which types of customers are driving your growth, which marketing channels are bringing in the most valuable users, and which product features are most popular with different customer groups.
Look at the arrows next to the chart below to see the direction on how to analyze the evolution of cohorts, within one cohort (left to right) and across cohorts (bottom-up) over time. This will have a substantial impact on your strategy. This example focuses on the number of times customers use your product on average per month.
As your product and acquisition channels evolve, cohort engagement and retention should increase over time. You may even see a bit of a smile in some user cohorts. This is upselling. As users are getting to know you better, engagement, and with it revenues, often increase over time. This should be the case for enterprise plays, and it sometimes happens for consumer businesses as well.
Predicting cohort behavior in the early days.Beyond analyzing past data points, cohorts also give you a good indication of future performance. Past engagement is the most correlated with future retention. Therefore, entrepreneurs usually focus on usage to predict future performance, and detect ‘hidden churn.' Hidden churn is when customers are inactive but still paying. Inactive customers will eventually leave, so you should monitor the health of your customers, quantify inactive ones, and try to re-engage with customers you might be losing.
If we go back to the first example in this post, here are some observations from the founder on this topic: ‘Is retention of users the same as retention of revenue? It is similar. There is the structure of our MAU. In January, we had 62 million MAU, and just 26 million users installed Flo in the last 12 months. 36 million users (60%) installed Flo over a year ago.’
Is it better to look at retention or churn?When talking about logo/client retention, churn makes the most sense. When looking at revenue/dollars, retention is the most straightforward. One is the opposite of the other, so it doesn’t really matter.
Think about this. Your customer churn rate can be very different from your revenue churn rate. For example, if you lose several smaller customers but retain your large ones, your customer churn might be high, but your revenue churn remains low. Customer churn cannot be negative. Revenue churn is negative if the upsold revenues from your retained customers are larger than the churned revenues. Ultimately, revenue retention matters more than customer retention, but you want to track both to have a full picture.
Sometimes cohorts are not enough.While cohorts give a nice granular view of customer groups, I usually suggest my founder go even more granular and focus on specific great clients. This is especially relevant when a few customers are developing above average. Here is an example of a great company with a PLG product I looked at that sold a productivity app to enterprises. Their core metric was the number of users within a specific company using and paying for their product. It allowed them to ‘zoom in’ and learn from those clients. But also to predict the growth potential of others and cohorts in general.
Last week, my friend Jan from Credo Ventures visited us in our office in Luxembourg. These spontaneous catch-ups are usually the best ones. We often have friends from the industry visiting that happen to be in town. Do not hesitate to let us know; we'd love to host you.
Life is awesome,
Yannick
European VC Europe