Mobile App Analytics

Ask any mobile business worth its salt where they stand on the subject of analytics, and specifically mobile app analytics, and I can guarantee you’ll hear some pretty positive noises. That stands to reason. Mobile is a digital business, and digital businesses are able to collect data, and allow that data to drive their decision making.

But when you ask exactly what form those ‘analytics’ might take you’ll get a whole range of responses. It’s much trickier to really define exactly what we expect from analytics - and it is a term that covers almost everything from the highest-level metrics such as daily active users (DAU) to the cluster analysis and automated segmentation of the user base (to give just one example).

So what do you actually NEED from your analytics? In the most neutral spirit possible, I’d suggest the following five requirements are both non-negotiable and - for almost all app businesses - pretty much everything you require to build a truly data driven organization.  

Cover Your KPIs

Let’s start with the basics. You need to take the heartbeat of your app, and you need to be able to report success or failure at the highest level. That means a clear view on daily active users (DAU), engagement metrics like average session length and number of sessions per day, retention stats (how many users remain active in you app) and revenue metrics - if you collect revenue.

You’ll need to know where you stand on all these KPIs, and you'll need to be able to observe trends. It’s how you measure success, but also the first stage in diagnosing any challenges. You can't do without it (not that many would suggest you try).  

Segment (and report) By Event

It’s important to get a segmented view of user behavior. And it’s really important to be able to target specific groups of users rather than have all your marketing be of the ‘spray and pray’ variety.

But what form that segmentation takes is another matter. At a basic level many analytics packages will give you the ability to draw a distinction between users on different devices, on different app versions, and perhaps (if you have the data available) demographic criteria such as age or gender. That isn't enough.

It’s absolutely essential that you can both report and target against segments that are defined according to user actions and behaviors, and that means being able to segment against events. In this context, an ‘event’ is any user action. That means we can create segments (for both reporting and targeting) such as “nervous window shoppers” defined as “users who have entered the in-app store at least 5 times, and backed out from a purchase screen at least three times”.

Analytics should help you see all those KPIs mentioned above for precisely the user groups you define as important, and allow you to define those groups with the maximum flexibility possible. Then you are truly tracking what counts.  

Identify Opportunities With Funnels And Cohorts

At Swrve, we’re firm believers that analytics falls into the “it’s not how much you’ve got, it’s what you do with it” bracket. Yes, it’s great to have big data at your service, and to be able to spend time identifying the patterns and trends that shed real light on your business. But it’s even more important to be able to influence outcomes. Ultimately, analytics needs to be about informing action.

On that basis, it’s important that any mobile app analytics solution includes those reports that help identify opportunities and suggest potential campaigns. One great example is the funnel analysis - in which ‘fall off’ can be identified in any sequence of related user events. Another is a cohort analysis, in which key metrics (such as retention, new users, engagement etc) are reported on a cohorted by day basis.

Both these analyses enable the organization to see where things are going wrong, and where they are going right - and isolate those particular aspects of the user experience that may need more attention.  

Measure Campaign Success

We too often think of analytics as relating solely to the metrics associated with the app itself. That would be a mistake. Equally important is carefully tracking and measuring the effectiveness of all app-based campaigns - a category that includes changes to user experience, the display of in-app messages to target groups (and in specific locations), and push notification campaigns.

An analytics solution should be capable of not just reporting on basic metrics of success such as click-through rates and user behaviors (such as purchases or social shares) influenced by these campaigns. It should also be able to deliver a holistic appraisal in terms of every KPI that counts. In this way analytics is not only tied to action on the ‘supply’ side (as above) - it is also used to establish just how effective that action was.  

Support Mobile A/B Testing

A further extension of this last point relates to A/B testing. As a practice, A/B testing on mobile is increasingly popular, for reasons that we hardly need to go into here. We’ve discussed them long enough elsewhere on this blog after all!

But where does analytics come into this equation? Quite simply, it provides the context and most important measurement behind real A/B testing, and most specifically it enables test results to be calculated with real statistical significance

That last point matters, because a ‘winning’ variant in an A/B test is not merely the variant that delivered more purchases, or greater engagement. Natural variation between groups means that unfortunately such a basic method of evaluation is pretty much worthless. On the other hand, a good analytics package will enable you to truly determine which variant is the winner in any given test - and also allow you to track and measure the effects of any test variants on all KPIs in your app.