We’ve gathered a couple of interesting cases from FraudScore clients about the Event Report and its efficiency in campaign analysis.
Let’s start with a reminder that FraudScore uses a four-risk level approach to traffic quality evaluation:
The FraudScore team always recommends paying attention to traffic with a FraudScore of more than 23. For that reason, starting from the blue zone, managers might need to dig deeper and analyze campaign traffic.
Case 1: Direct Advertiser
In this case, the Event Report helped achieve a full picture of fraud, including a combination of click spam and bot traffic, all sent by one affiliate.
The advertiser noticed that TTI distribution was out of the ordinary and looked suspicious. Thus, the FraudScore team decided to dig deeper and help the advertiser drill down the fraud report.
Our team customized the Event report and combined TTI parameter and Events:
Bot traffic was the main fraud type used by this affiliate. Usually, a low TTI and no events are symptoms of bot traffic. In this case, 2% of conversions had a very low TTI. However, the majority of conversions had normal TTI (e.g., 1–5 minutes and 5–15 minutes), which is a symptom of good traffic.
If we take a look at the combination of both TTI parameters and Events in the FraudScore report, we can see that there were no events for conversions with good TTI. Moreover, while comparing this source with other affiliates, our team noticed that this described behaviour was unusual. Other affiliates’ traffic had conversions. So, the conclusion was that this affiliate had bot traffic.
Additionally, FraudScore detected various other symptoms of bots, including old operating systems or browsers being used. In fact, 98.4% of traffic was sent from iOS 9.2, whereas the current version of iOS is 14.6.
Moreover, if we take a look at the conversions that were marked with click spamming (i.e., very high TTI), we see that they have real events. It looks like these conversions are a form of stolen organic traffic — the users are real. They were most likely added by this affiliate in order to trick the antifraud system and show more good conversions with in-app events.
As a result of the analysis, FraudScore’s team gave following recommendations to the client:
In two simple clicks, we were able to set the Event Report according to our goals, and we helped the advertiser understand why the traffic from this particular affiliate was not trustworthy. It took less than 10 minutes to see the whole picture.
Case 2: Mobile Developer
Another advertiser case — one affiliate had his traffic in the green and blue zones — Good Conversions and Low Index Fraud Conversions. But TTI was alarming — too low to be real. FraudScore’s team decided to take a closer look.
By using the Event report, the team dug deeper into the events from conversions with low TTI. 90% of users from these conversions finished registration, but didn’t proceed further and didn’t realize any other event:
This looked suspicious. In order to compare, the team examined traffic from another affiliate and the event report there. It showed that 30–40% of users proceeded further after registration, and realized other actions (i.e., events).
This comparison demonstrated that our first affiliate was not being honest, and his traffic was fraudulent. It appeared that low-TTI conversions were done by bots, which is why there were no events after registration. This stood out from the regular user actions when comparing this affiliate with the other .
In this case, the Event Report not only helped detect fraudulent sources, but assisted in building a strong argument for the case. The comparison with the other affiliate was used as proof of irregular activity and fraudulent traffic.
FraudScore’s recommendations were to reject Red and Orange Fraud Index Traffic, and to reject all conversions with Low TTI symptom — 693 conversions out of 1079 total from this source (around 64% of traffic sent in a month).
The FraudScore team is always there to assist and consult our clients. We help read reports, understand detected ad fraud, and build fraud cases. You can always send us your cases and questions, and we will help dig deeper into traffic quality analysis! Contact us via sales@fraudscore.mobi or through your personal manager from the FraudScore team.