10-Step Approach to Fighting Fraud

By Diksha Sahni | July 17th, 2017

Mobile fraud is ubiquitous and, to some degree, unavoidable. But advertisers can minimize the risks by becoming more aware of the target KPIs of the app and working closely with the ad networks, publishers and attribution partners.

We have outlined a ten-step approach as a starting point for advertisers to help fight fraud:

1. Understand the target KPIs of the app: Advertisers can play an integral role in early fraud detection by having a process to monitor traffic and KPIs to detect any suspicious activity.

2. Background analysis: Know where the risks are. Make sure you understand the different types of fraud and how to detect them. Diversify the sources of information — networks, media sources, fraud detection companies, and publishers are all your friends in staying informed about how to identify fraud. Do not rely on one source; make sure to get the full picture to educate yourself on not just identifying fraud, but also interpreting heuristics to detect fraud.

3. Incentives and management structures: Set the highest standards of quality with your UA managers to focus not only on bringing installs, but legit installs. Incentivizing UA managers only on volume may push them to close their eyes on fraud.

4. Work with attribution and tracking partners: If you do not have resources to tackle fraud internally, work with a fraud solution offered by main measurement partners such as AppsFlyer, Kochava, TUNE, and Adjust.

5. Set up an internal infrastructure: Build a scalable internal infrastructure with your Business Intelligence teams to easily and quickly understand and analyze patterns. By having a scalable infrastructure (a combination of technology and people) that can process the huge amount of campaign data coming in, you can help to take timely actions to fight and prevent fraud at various levels.

6. Select your sources: Before beginning campaigns, advertisers must ask their partners about their approach to fraud and choose businesses that have robust prevention measures in place. Be sure to work with only those partners who take the most proactive stand against fraud.

7. Implement guidelines and target metrics: Set up a process to implement campaign management guidelines with all your supply sources. Be careful, though, as you might miss inventory opportunities if you inaccurately define target metrics or range.

8. Measure continuously: Once you have developed an understanding on what patterns can be used to identify different types of fraud, build a set of data and calculate metrics based on patterns across all different inventory sources. Define corridors for your target metrics and each pattern that you will be focusing on based on the analysis. Take into account individual factors of your app that may lead to different empirics than general market averages.

9. Exception management: The amount of data being generated is huge. Set up a system in which certain patterns trigger alerts or actions automatically. For ad publishers and DSPs, monitoring traffic and campaign-related KPIs for suspicious patterns is of critical importance. For example, if you observe an unusually high app-install rate, that the time between click and app install is uniform for every user from the same traffic source, or that the time between click and install is inexplicably short, that should raise a red flag.

10. Implement a feedback loop and incorporate learnings: Be proactive. Should you identify a worrisome traffic pattern before your supplier, share as much data as possible so that your partners can learn from the incident.

Find out more about identifying and fighting mobile ad fraud with our latest eBook! Download your free copy here!

Diksha Sahni
Diksha is a Content Marketing Manager at AppLift and is based out of our Bangalore office. When she is not behind her computer writing, you can find her binge watching her favorite movies, finding her happy place at a dance studio, and checking off places on her bucket list.

12063761 1030943930291273 1184227129633473729 n