How To Spot Fraud In Your Mobile User Acquisition Stack

With worldwide mobile ad revenue on track to hit $75 billion in 2018, it’s not hard to understand why mobile ad fraud has been making SO MANY headlines in the business press. Pundits and professionals claim that fraud rates are worse now than they’ve ever been after climbing 30-50% since this time last year. Exact numbers vary from source to source, but it’s been posited that as much as 33% of all digital ad traffic could be fraudulent.

All this bad press in major publications has turned fraud into something of a boogeyman in the mobile marketing industry, unfortunately leading many publishers to misdiagnose data discrepancies, calling it fraud when in fact that’s rarely the case.

It has been our experience that 95% of what most mobile marketers end up blaming on fraud can be traced back either to human error or failures of process. In fact, most mobile publishers aren’t ever exposed to fraudulent activity until they set their install targets into the tens-of-millions, forcing them to expand their growth operations onto specialized affiliate networks, where less rigorous inventory vetting processes allow fraudsters to thrive.

In order to take a more dispassionate look at their data-driven marketing challenges, marketers should ask themselves a few key questions before considering the possibility that fraudsters have taken up residence in their growth stack:

Are You Working Exclusively With Premium Networks?
Facebook, Google, Pinterest and Snapchat all hold their advertisers to extremely high security standards. It’s extremely unlikely that a disparity between dashboards would be caused by fraudulent activity when exclusively working with these premium partners. Their processes have been refined to the point that it’s almost impossible to inject fraud at any point in the stack, making them an ideal choice for anyone just getting started with user acquisition. If you are using affiliate networks, there’s a greater possibility that you are dealing with fraud, but it’s still far from guaranteed.

Have APIs Changed?
Conduct a diligent audit of the APIs used in your growth stack and make sure there haven’t been any critical updates or changes made in the past few months. Some of the most popular APIs can change 3-5 times a year, and while providers may do their best to keep users abreast of updates, it’s entirely possible that a critical communication fell between the cracks.

Were Any Updates Delayed Due To ETL Issues?
Databases aren’t perfect. Issues in the extraction, transformation, and loading of information can lead to imperfect aggregations and inaccurate dashboards. Check with your engineering team to see whether or not any stalled database functions could be responsible for missing data.

Has Your Data Collection Been Setup Correctly?
One of the most common reasons for data discrepancies is the misattribution of data due to human error when setting up tracking links, data definitions and/or matching processes. Spend some time going back through your management dashboards and integrations to confirm that you’ve pulled the appropriate codes and keys so that your campaign data is being sent to the right place and includes all the parameters it should.

If, after all that, you’re still concerned that fraud may be affecting your UA campaigns, there are a few relatively easy ways of identifying which networks are responsible assuming you have access to user level data.

As stated earlier, marketers can be almost 100% certain that installs driven by premium networks like Facebook and Google aren’t fraudulent. This provides marketers with a valuable baseline that they can use to evaluate other networks. Start by identifying trends and commonalities in metrics like time-to-install, operating system version, and post-install event funnels among users coming from premium networks. Then, compare this data against the users coming from the networks you suspect may be driving fraudulent installs. Anomalies at the lower end of the spectrum could suggest you’ve been the victim of attribution fraud, whereas higher end anomalies point towards bot traffic.

Either way, if after a thorough investigation you’re fairly certain you’re paying for fake installs, use the data supplied by these methods to make an argument to the relevant ad networks as to why you won’t be paying your invoice. Only by putting pressure on ad networks to introduce stricter vetting practices can the mobile industry combat fraud and improve performance for publishers everywhere.

At Tenjin, we build the growth infrastructures that mobile marketers need to run effective, fraud-free user acquisition projects. Our user-level data warehousing solutions afford marketers the direct access to data they need to identify fraud quickly and confidently. To learn more, feel free to get in touch with us.

comments powered by Disqus

Save Money on your Mobile Marketing. Seriously.

Subscribe

We'll keep you in the loop with product updates and interesting content around data, APIs and analytics.