What is retention analysis?
Retention analysis helps product teams answer the question, “How many of our new users come back to the product?”
Sometimes referred to as survival analysis, retention analysis runs alongside your new-user acquisition metric to determine the percentage of user growth that turns into a recurring and profitable customer base.
Why is retention analysis important?
When you first get started, user acquisition metrics are exciting, straightforward, and appear to be top-of-mind for everyone from executives to investors.
However, focusing too much on new user growth without insight into why your customers decide to stick around can waste valuable acquisition resources and conceal key indicators for improving your customer retention. If the marketing team is paying heavily to acquire new users who quickly churn, those users’ customer lifetime value (LTV) might be lower than their cost of acquisition. If the company maintains steady acquisition spend that is higher than the customer LTV, the app will not survive.
Product teams need retention analysis to better understand how they can keep more of their users and answer questions such as:
- How long do various customer personas stick around?
- How long does it take a new user to come back to the product?
- Have recent product changes encouraged more users to return?
- What changes have negatively impacted retention?
- What changes are likely to improve retention?
- What is my churn rate?
By answering these questions and increasing retention, product teams can plug the holes in a leaky retention bucket to make their product more profitable.
What do you need to get started?
In-app user behavior
Product teams must know what’s happening within their app. That means tracking users as individuals or as groups as they engage in activities such as downloads, account creation, sign-ups, payments, downloads, shares, and more. They can also track each action back to its acquisition source — such as an ad campaign — in order to evaluate which events and sources are best correlated with retention.
Retention can be calculated with any action that a user takes (like logging in), or you may choose to measure only those actions you consider valuable, such as converting or sharing.
Next, product teams need to define what form of retention is meaningful to their product. Start by defining a goal event that is specific to your app, and the number of times that a user is expected to come back to complete it. For a social media app, the retention strategy may look like a user returning to their feed within three days of their last login. For a banking app, the strategy may be across a longer time period — such as expecting a user to return to the app once every few weeks.
Sample definition: A new customer has been retained if they’ve taken any action within the first 20 days. If they don’t take action for 20 days, they’re considered to have churned.
Product teams can divide their users into cohorts so they can compare which cohorts demonstrate the strongest retention. In some cases, cohorts are defined by the week in which users created a new account. For example, each month would have four cohorts, one for each week. Product teams could compare January week one to January week two or to February week one to see if retention has improved over time. Alternatively, cohorts can be segmented according to defining properties such as attribution source, location, plan details, etc.
Product teams can compute a variety of different retention rates that best reflect the goals of their business.
Retrospective retention analysis answers the question: “How many of our new users remain customers?” It calculates how many members of each cohort were retained over a trailing period of time.
For example: After 30 days, 28% of users that signed up for the app on January 1st were retained.
For a consumer music app, that might mean calculating retention for all users at seven days, 14 days, and 30 days. For a marketing automation software with long contracts, that might mean calculating year-over-year.
Product teams can also calculate retention by segmenting their data by customer cohorts. When evaluated in conjunction with the LTV of each persona, the results can be illuminating. Some personas may demonstrate high potential but have low LTV because of their low retention numbers. By increasing their retention, teams can unlock much greater value.
And finally, product teams can evaluate retention based on acquisition sources. It’s not uncommon for certain sources to produce customers with either very high or very low retention rates. Ads, for example, might bring in a glut of new users who quickly lose interest. Social shares, on the other hand, might surface users who were dying for a solution and who are much more loyal.
How do retention metrics trend?
Just as important as measuring retention is measuring it over time. Knowing that a product has a five percent 30-day retention rate is much more meaningful if it’s compared to previous rates. If last month was six percent and the month before eight percent, product teams have a leak to fill. On the flip side, if it’s up from four percent, that’s cause for celebration.
Is there enough engagement behind our retention?
Product teams can use engagement to measure the quality of their retention. By choosing engagement events that are most meaningful to your product, you can determine whether users simply sign in and out, or if a user logs in and spends three hours commenting on posts. By measuring engagement alongside retention, product teams can understand more about the value users are getting from your product.
Product teams can isolate the variables that appear more highly correlated with retention — such as referring friends, saving trip details, creating a report, or making a purchase — and then focus their product design and marketing efforts on driving users to attain those early achievements.