Retention rate is one of the most critical metrics for early stage startups. A company with low retention is like a bucket full of holes, you'll need to work very hard to fill it and even then you'll most likely fail.
In this post I'm going to share with you everything you need to know about retention rate, including how to calculate your retention rate, how exactly retention rate is defined and more.
Retention rate is one of the best indicators that a company has built a product or service which is providing value.
Think about your favorite burger joint. Perhaps it's a long drive across town but you go back frequently non the less, why? You go back because it provides you with some much value that you go out of your way to experience their product. This hypothetical burger join has a high retention rate when it comes to you as a customer and since you're an average consumer, it most likely has high retention among the majority of its' customers.
Now there are exceptions to the rule. Not all places of business or online services require you to visit them frequently in order for you to benefit from their services. You might go back year after year to the same dentist, or use the same travel agent, but those visits are infrequent. Volume is relative and when it comes to measuring retention rate, you'll need to keep this in mind.
A product with high retention is also much more likely to have product-market fit than a product with low retention.
Retention rate doesn't have a universal definition. In layman's terms retention rate is the rate at which you retain your users or customers. Retain in this context means the percentage of users or customers which return to your service or use your product within a defined time period.
A time period which is often used to define retention rate is weekly retention. A user which returns to an application at least once a week for 6 weeks after their first use would have a 100% weekly retention rate for the first 6 weeks of their lifecycle. Since time is never ending, you need to define retention rate both in terms of period grouping and duration.
For example, you could look at daily retention for the first 30 days, or monthly retention for the first year.
Another way that some company's measure retention rate is the number of unique days of usage. Say for example that within the first 30 days from signup someone uses the application on 7 different days. You could say that this user has a retention rate of 7 unique days of usage within the first 30 days.
Wearable device company's or services which have by design intermittent usage may prefer this method. The advantage is that if you see a change in behavior once a user has used the service X times, then you'll want to track the percentage of your cohorts which reach this milestone.
Before you can calculate retention rate you'll need the following:
Once you've defined retention rate and have enough data then you'll want to go with a classic cohort analysis.
The screenshot above shows an example of a retention-based cohort visualization. In the example above, the rows represent weekly cohorts of users that signed up in that given week, while your columns represent week count from signup.
Since everyone visits the application the same week they signup, column 1 is 100%. The subsequent cells represent the percentage of each row which revisits the application (or reuses the product) as time goes on.
We can see that by week 7, only a fraction of the users are still returning to the application.
Once you're done building the cohort table, you can calculate the averages per period and turn it into a retention curve.
As you can see in the screenshot above, the curve flattens out from day 7. You'll want to see a similar pattern since this indicates that a certain percentage of your users are sticking around indefinitely.
From a technical standpoint, you'll want to create a data set which has a row of data per period grouping per user. You'll then want to include a column of first usage or signup per user across every data point. This will then allow you to pivot the data and group users into cohorts. You may also want to include other interesting dimensions such as age, gender, lifetime value etc so you can calculate your retention rate by different user properties.
Most startups don't save historical data on visits to their application. In order to accurately measure retention rate, you'll need historical data. A tool like Segment is a great option since it makes it easy to record historical data in your own database.
If your retention rate drops to zero (or gets close to it) at a certain point in your cohort analysis then it means no one is getting enough value from your product to stick around. You'll want to go back to the drawing board and identify why users are not sticking around.
If your retention rate hits zero early in the user's life then you'll either want to do significant iterations on the product, or reassess the concept as a whole.
If your retention rate slowly moves towards zero after many weeks, you'll want to double down on user research and one-on-one interviews to better understand why users aren't using the product or service in the long-run.
If you're trying to raise investment for your company you've most likely been asked to share your user retention with potential investors. Investors love the retention rate metric because a service or product with high retention will be lower risk than a service or product with low retention.
Every startup should understand their retention. There are many different ways to calculate retention and a data-driven company should invest the time to segment their users in different ways to see who is being retained and who is fleeing the ship.
Once you've determined your retention rate the hard work begins in moving the curve further up the graph.