With Google I/O just around the corner, we again find ourselves in a heated Android vs. iOS debate. And here at Swrve, debates turn into digging, and digging often turns into some pretty awesome data… which of course, we had to share.

Our latest analysis (details of the methodology are below) found that almost one in five in-app purchases on Android smart devices are in fact fraudulent transactions delivering no revenue to the app developer. A full 19% of all purchases are fraudulent, which both impacts total revenue, but perhaps more importantly can lead to poor decision-making in data-driven organizations unable to filter out the counterfeit “noise.” When running in-app marketing campaigns or A/B testing user experience, revenue numbers reported may be wildly inaccurate and thus suggest ROI is delivered when it may not be the case. Which as you can imagine is something of a problem.

More anecdotally, we've seen instances in which for limited time periods, up to 90% of all revenue events by value are fraudulent - a situation in which it becomes almost impossible to gather clean revenue data in most analytics and marketing platforms.

If you haven't had the chance yet, you should check out our Mobile OS Report that we released last month. It looks at the differences between iOS and Android users – from retention to engagement and monetization. The report revealed that on average, iOS users spend 45% more on in-app purchases than those on Android, with iOS users who pay at all spending 10% more and iOS users also 32% more likely to make any purchase at all.


The 19% figure is based on all in-app purchase (IAP) data on Android platforms in the month of May. The analysis was limited solely to apps that include in-app purchases as an option, with IAPs being strictly defined as those fulfilled via the Google Play store. This still left over 50 apps in the analysis and the revenue events cover millions of users. Using Swrve's 'fraud filter' technology all local revenue events were verified against Google Play receipts and thus determined as valid or invalid (or fraudulent).