User Retention: How to Choose Your Metrics with Your Analytics

Mar 18, 2019

It is one thing to attract a new visitor or user to your website. It is quite another to have them stay there, explore, make a purchase, and then come back for more in the future.

The first step is “acquisition.” It’s is the process used to acquire that visitor to your site for the first time. But once they’re there, you want them to engage with the site and come back for more. The latter is “retention.” It’s great that you were able to acquire the visitor, but a visitor is only as useful as your ability to keep them. Only then will that user be able to bring long term value.

Acquisition vs. Retention – Which Brings More Profit?

In the typical brick and mortar store, acquisition begins the moment they make their first purchase. Then these stores try to offer enough value that they are able to maintain those customers in the future.

In an article published in the Harvard Business Review, it is estimated that customer acquisition costs about 5x to 25x the amount of retention, and that increasing that retention rate by only 5% can boost profits by anywhere from 25% to nearly double. Customer retention is extremely profitable.

How to Measure Retention Online

Offline, measuring customer retention involves determining how many customers came back for another purchase. Though it may be difficult to set up a way of measuring it (for example, if a customer pays in cash and does not give their name), those that are able to measure customer retention do so by seeing which are still coming to the store for future purchases. Those are their retained customers.

Online, however, defining customer retention may be a bit more personal. It is something that has to be unique to the goals that you’re trying to achieve.

For example, let’s say you have a mobile app that makes its money off of paid advertising. Your idea of retention may be how long a person plays, or it may be how often they open the app. But another website that sells a product may measure retention in terms of repeat business.

Other metrics also may mean more to one type of online business than another:

  • Some businesses expect frequent purchases, so a customer that comes back once per year may not truly be a “retained user.” Other businesses may use retention analytics that are not bounded by time, and simply refer to customers that come back for further information.
  • Some businesses may look at usage, even if purchases are not made. For example, Facebook looks at monthly active users and engagement metrics to determine retention.

Every website or app has their own definition of what retention means to their business, which is why it is important to have analytics software that can adapt to these different methods of retention, and one that can be easily altered depending on the goals you have for your organization.

With that in mind, there are a few ways that we can define user retention that are consistent across multiple businesses, and all made possible via your app analytics software."

Classic Retention (aka N-Day Retention)

Classic retention looks at the number/percentage of total new users that returned to use the app or website again on day X (referred to in this case as day “N”) after their first visit.

This type of retention is meant to provide the business with a percentage that reflects the website’s dropoff rate over time after their first visit. A typical chart may look like this:

Classic retention is best used by applications and websites where daily engagement is not only a goal – it is also a possibility.

A website that sells clothes, for example, probably would not benefit from N-Day retention because there is limited likelihood that a user will come back to the site every day for a long period of time.

But with a mobile app, game, or other website where daily engagement is a possibility with the right functionality, classic retention can be a simple but beneficial way to monitor website usage over time.

Bracket Retention or Range Retention

Different apps have different usage patterns. One of the challenges with N-Day Retention is that usage rate can vary wildly for some websites or apps, and if you are too focused on an exact usage amount, you may be over-eager to make changes that are otherwise not meaningful.

That is why a more accurate measure may be “Bracket Retention,” also known as “Range Retention.” This type of retention analytics measurement focuses less on which day a user returns, and instead calculates whether or not a user returned in a pre-defined range.

Businesses that expect a standard amount of time to pass in between visits may be especially interested in bracket retention. For example, if you operate a women’s clothing eCommerce site, you may not expect your users to come back or make a purchase every day, but perhaps you are Interested in determining how many come back for each season – ie, every 3 months.

Bracket retention makes intuitive sense for any app or website where daily recurring visits is unlikely, or where the goal is visits at rough intervals. It also may provide you with a more accurate measure of usage, and takes some noise out of the data by grouping users together.

Rolling Retention or Unbound Retention

The above retention analytics are focused mainly on time – did a user come back at a specific period of time. Time always plays a factor in retention, but for many businesses, “time” only tells part of the story.

Often the real question is “did the user come back at all.” Many businesses fall under this category, from rideshare services to booking services to eCommerce to casual gaming apps. There isn’t necessarily a guaranteed range of time that a person may use the service, nor is it always in the best interests of the company to push people into speeding up their purchase rate.

No, the question that many are asking is simply whether they came back at all, and that is where analytics like “Rolling Retention” come into play.

Rolling retention doesn’t necessarily care when a user came back – at least not in terms of a specific date. What it cares about is whether they came back at all. The frequency of their return still plays a role in the calculation, but the focus is on whether or not they come back any time within a meaningful time frame, rather than at a specific point in time.

To understand what this looks like:

Pretend that you sell camera equipment, and on February 1st, 2019, there were 20 purchases from new users. You want to make sure that those users were satisfied enough to come back and become recurring customers.

But camera gear is not something a person buys every day. It is not even something a person buys every week. It is something that is purchased only when needed, and that may be February 8th, or May 15th, or January of 2023.

So the question is not when they come back, but if they come back, and that is where unbounded retention can be so beneficial. As long as the user comes back, it counts as a retained user, which makes sense intuitively for many types of industries.

In addition, rolling retention can be re-evaluated at different intervals.

You can have a rolling retention rate starting at day 0 (the day of the first purchase), and you can have an additional rolling retention rate starting at a different day, such as day 14 – thus eliminating those users that may have used the app for multiple days in row before dropping out.

An example of how this may be implemented is in gaming. There are several “build” games, where a user builds an army or a city and has to wait in real time for the builds to finish. At first, the game tends to be pretty fast paced, with a variety of builds and upgrades that take only a few minutes to complete.

But as the game goes on, the builds take longer. It is not uncommon for these games to have considerable engagement for the first several days as their play is reinforced by rewards. But the question for app developers is whether their game is still engaging enough to continue to draw in play once those rewards start to diminish in speed and quantity.

Rolling retention rate is a useful way to calculate retention when specific days or time intervals makes less sense for the business.

Which Retention Rate Calculation is Best?

One of the advantages of using a more advanced analytics product beyond the freemium products like Google Analytics is that the data can often be very easy to analyze, and requires minimal data background. Artificial intelligence is also being integrated more into some of the newest and most modern analytics software packages. The products that use AI (like Air360) are able to analyze data automatically for you based on the information it collects. That makes it possible to explore your retention rate automatically, and spot opportunities with minimal work from you.

The choice of which retention rate calculation to use should be based on the type of business you operate and how your users make purchases. You can always track more than one type of retention as well and see how that compares to revenue, or to gauge if there is consistently for any of the measurements.

Determining Which Type of Retention Measurement is Best for You

Here is a cheat sheet for understanding which one is best for which setting:

N-Day Retention:

  • Best for apps or websites that can easily be used daily.
  • Best for engagement and log-in metrics, not (necessarily) purchase based.
  • Very easy to understand.
  • May not accurately reflect retention for most businesses.
  • May be prone to hiccups in the data, such as when use of the app coincides with a holiday.

Range Retention:

  • Reduces the noise of N-day retention.
  • Can be customized to your likely return user rate.
  • Better for establishing and monitoring trends.
  • May hide important data by grouping all dates together.
  • May still be prone to some noise.
  • May not be right for all app/website types.

Rolling Retention:

  • Measures retention in a broad, easy to digest way.
  • Isn’t bound by specific dates and times.
  • Customizable based on what you’re trying to measure.
  • Equates all time periods equally, as a daily user = a user that comes back after 5 years.
  • May make long term analysis more challenging.
  • Data changes often as users come back after long periods of time.

Keep in mind that these are not the only ways to measure retention either. There is also a form of bracket range retention, where app users look to see if someone comes back within a specific time period after they first use the app. There is also retention based on whether or not the user unsubscribed or closed an account, and how long after initial signup did that occur.

But above examples are the most common, and can provide you with valuable feedback for your website or mobile app.

So You Have Your Retention Metrics – Now What?

The moment you use your analytics software, you should start determining which retention type(s) you want to use and start tracking. That is because what makes the data on retention so valuable is that it can be used in the future to compare against when you make changes with your app or website.

For example, if you notice that retention appears too low on your website, then you can determine if there is an action you can take (ie, specials, email marketing, new website design, new features, etc.) that may attract them back and keep them there in the long term.

You can also start determining new ways to address retention to see if you can improve upon those numbers. The most common way is to simply offer a better product. For apps, that may involve reducing bugs and improving functionality. For websites like eCommerce, that may include faster shipping or unlimited returns.

Other examples include:

  • Seamless Transitions – Make it easy to sign up, easy to make purchases, and easy to engage with the app without the extra steps that may turn off users from continuing onward.
  • Generous Time Rewards – Make it more “profitable” for the user to log into the website or app more often. In games, for example, that may mean offering generous rewards.
  • Utilize Push Notifications/Emails – Using push notifications and email newsletters strategically can be useful for improving engagement. Just be sure that the user benefits.
  • Add a Social Component – When communities are built upon a website or app, the users that interact with those communities may be less likely to abandon it.
  • Random Specials – Providing liberal specials at random times may encourage users to come back often to see if the special is worthwhile.

Your product or service will guide these choices, as daily gaming apps may have very different retention and engagement strategies than, for example, an appliance review website. It’s also okay to get it wrong.

Remember, once you have an analytics system in place, you’ll be able to collect and review an almost unlimited amount of data. You’ll be able to see what is working and what isn’t. You’ll be able to change paths because the data will always be there. Because you will have collected so much data, you can change strategies or adapt in the future if you make a mistake.

Nevertheless, there are always ways to improve, and compiling these data with a powerful analytics tool is a great place to start. The more you can learn about retention – as well as the rest of your website’s performance – the more opportunity you will have to find opportunities and make changes that will be capable of making measurable improvements to revenue and growth.

Try the Air360 analytics program for yourself by contacting us on our main website at