The principal aim of any app marketer is to acquire new users and retain them over time.
As we all know, keeping users engaged is a good benchmark for understanding the success of your app, and also results in higher returns when it comes to revenue.
However, churn is an unavoidable part of the equation. There will always be a proportion of users who leave an app, and that number is high, too: in fact, we estimate that almost 80% of users have churned the day after an app install. Skip to day seven, and only 12% of users are still active.
But does that mean the rest are lost forever - with no hope of opening your app again?
Well, no, as it turns out. As we’ve seen before, many churned users are likely to come back to an app, whether that’s a couple of months or up to a year after their last session.
Today, we’re looking at this theory in more detail - and drilling down per vertical, to see how returning users stack up across different sectors.
Do users return after long periods of inactivity?
To find out, we analyzed data over a year period across eight verticals. The results show the average rate at which churned users will come back, defined as returning after a gap of two or more months, where the user didn’t register a session. It could be that they kept the app on their phone without opening it, or that they uninstalled and then reinstalled the app.
The results were encouraging: the number of users returning after a two-month break was equivalent to 17% of all new installs in that period. After three months, an estimated 11% of churned users will return - and even further down the pipeline, at six months, that number stands at 4%.
Because churn rates vary across verticals, it’s natural they’ll also have different rates of returning users. Let’s take a look at how these differ. We’ve split the verticals into two different graphs, according to return rates above and below the median.
E-commerce’s rates are especially high. Three months after the last session, the return rate stands at 18%. This isn’t surprising, as e-commerce apps aren’t as conducive to casual browsing or regular interaction as say, social apps. Often, users only open the app with a specific need - they want to search for or order a product. Unless you’re shopping in particularly high volumes, this is likely to be every few months rather than every few weeks.
On the other hand, it’s important to note that many consumers are daily drivers - which is why e-commerce apps retain so well in the first place.
Utilities tell a similar story. Often, these apps serve one single purpose - take a scanner app, for example - and will only be opened for one particular use. That’s not usually a weekly or even monthly occurrence, but these apps are useful enough for the user to keep on their phone for the next time.
Games also have high numbers of returning users. While other verticals’ return rates gradually fall more or less in line with the average, here, the rate of churned users returning is consistently almost double that of the average. Games have a particular pulling power: half a year later, almost 8% of churned users will return. Even a year on, an average of 2.3% users come back. While that figure may seem small, when aggregated suddenly your retention analytics would be severely compromised.
Now let’s take a look at the lower part of the graph. While these verticals’ return rates fall under the median, it’s still really valuable to know when users are likely to return to your app: if not, you won’t have a full picture of the user lifecycle, and you won’t be retargeting as effectively as you could be.
Take Travel: the vertical still has noticeably high rates of return, with almost 8% of users returning after a three-month break. That’s not unusual when you consider that online travel agent, airline and hotel apps are the most popular types of travel apps.
These sorts of apps are, for the most part, purely functional - and often designed to serve seasonal, specific purposes, like booking a flight or hotel room, rather than for casual browsing. However, knowing when users tend to open a new session or when they reinstall makes a massive difference when it comes to retargeting: users who typically churn after three months could be targeted with tailored activities or offers to help draw them back in.
On the other hand, Social’s rates of returning users are noticeably low - just look at how closely the purple line touches the horizontal axis, right from the start. Even after two months, only 1.3% of churned users are likely to come back.
This makes sense: while social networks are the cornerstone of modern communication, they’re pretty all or nothing. Either users are engaged right from the start, or their interest dwindles immediately. But the potential pay off - of pulling in highly engaged users - means it’s critical social app marketers know when users typically churn or return, to have as strong a chance as possible at engagement.
Why we work with unlimited lookback windows
For the reasons outlined above, we realized pretty fast the importance of having unlimited lookback windows.
That's because sessions instigated by users who had already downloaded the app and were now returning after, say, three months, would actually be counted as a fresh install. Not only that, but they could also be attributed to a paid marketing campaign - costing the advertisers and messing with their data. If your user records don’t stretch back far enough, your install numbers would be off by around 11% - and you’d be paying ad networks for up to 11% of existing users that look like new acquisitions.
So, how much does all this misattribution cost?
If your user data gets thrown out after three months, how much could these discrepancies potentially cost? Let’s take the games vertical as an example. Say your app has around 2,000 new users per day, an average 26% of which comes from paid installs. At four months in, the amount of churned users who could come from paid channels - 60 or so - will open a new session. But instead of being counted as returning users, they’ll be seen as fresh installs, and could be attributed to a campaign.
Our benchmark tool puts the current cost per install at $2.26. That means that on a given day, you could be wasting almost $150 on misattribution. That may not seem like a lot, but in a single month, it comes to $4,500 - money that could be much better spent targeting new and existing power users.
Aside from the direct waste of money, misattribution has a knock-on effect on the accuracy of your data and subsequent re-attribution campaigns - but working with unlimited lookback windows and uninstall and reinstall tracking will help you gain as precise an overview of the user lifecycle as possible.