Fraud traffic or invalid traffic (IVT) or bot traffic is one of the biggest problems in online advertising industry.
We combined data from few respected sources to help you see full picture of what was happening in 2018 in app installs, programmatic display and video media, by countries, categories, OS, devices and fraud types. Some information is relevant only to 4th quarter of 2018.
Let’s start with global online ad expenses from 2015 to 2019.
Industry is growing 12-16% each year. But fraud grows even faster.
In Q4 there was a 24% increase in fraud traffic compared to Q3 of 2018.
Mobile in-app programmatic advertising had a 23% fraud rate in Q4.
Android devices had 1.7x more ad fraud compared to iOS.
Sites with ads.txt had 23% less fraud than sites without ads.txt.
Android devices are still more exposed to fraud than iOS, but the gap is getting smaller.
It is almost impossible to find an advertiser who is looking for app installs without hard KPIs.
Advertisers now require publishers to reach not only RR, but also in-app activity or sales. Incentive traffic is something that almost totally disappeared in CPI affiliate industry because of fraud involved in it.
Overall in-app programmatic advertising fraud increase almost 33% from Q3 to Q4.
Regarding in-app advertising fraud spread by OS, we can see this picture:
So, what kind of fraud is used?
Let’s have a look at some main types.
In Q4 2018 fraud assignment looked like this:
Incentivised traffic is not a fraud on its own, however using it, is one of the easiest ways to increase number of installs on a non-incent CPI campaign. Incent abuse can be detected by monitoring high conversion rates (CR). Unfortunately, using click spamming or click injections by frauders, makes detection more complicated.
Click injection occurs when a fraudulent app on a device monitors other app installs on the device and creates fake clicks while the install is in progress. It results in claiming the attribution of installs and generally click-to-install time (CTIT) is short.
Click spamming is a fraud type with simulation of high number of fake clicks from real devices to get the attribution of organic installs. Most of the installs have long click-to-install time (CTIT). Fraudulent click usually occur in two cases: when user sees your ad, but actually did not click on it or worse, when did not see your ad at all.
SDK Spoofing has become the most discussed fraud type recently. It involves manipulation of attribution tools. Fraudsters essentially create a bot, that hides in an app and generates clicks, installs and engagement signals to the Mobile Measurement Partner (MMP) fooling it into thinking the events are real.
In 2018 SDK spoofing was probably the fastest growing type of fraud.
Bots/Emulators. Similar to device farms, bots and emulators use fake devices inside desktop computers to click on ads and then install the app. Since the device is not real, it is even easier to reset devices to start over.
Device farms uses many real devices with unique IDs, operated by special software, which click on ads and convert them into installs then reset the devices to start over and over. Can provide as many clicks or installs as needed.
How do device farms look like? Here you go:
Let’s have a look at some more graphs!
Top 15 countries by fraud (all types of programmatic advertising):
We can see that he top five countries that are victims of ad fraud are India, Australia, Indonesia, Argentina and USA.
Fraud rates by device (programmatic display):
Fraud rates by device (programmatic video):
Fraud rates by impact of ads.txt:
ads.txt (Authorized Digital Sellers) is a file, which specifies list of other companies who is authorized to sell their inventory.
Android vs iOS display fraud rates:
Android vs iOS video fraud rates:
Android vs iOS smartphone only fraud rates:
Android vs iOS tablet only fraud rates:
Android vs iOS total fraud rates by devices (video and display):
Android phones application categories display fraud rates:
Android phones application categories video fraud rates:
iPhone application categories display fraud rates:
iPhone application categories video fraud rates:
Total iOS by category fraud rates:
Total Android by category fraud rates:
World Federation of Advertisers (WFA) forecast that by 2025 fraud in online advertising industry will become the largest market of organized crime and will be worth $50 billion, second only to the drugs trade.
Without sufficient counter measures, it’s easy to produce scenarios, where ad fraud revenues rise to $150 billion per annum in the same timeframe.
Much can be done by advertisers to improve the situation. Setting new standards, making contractual changes, demanding increased transparency and putting in place internal resource dedicated to countering ad fraud will slow down or maybe even localize fraud in certain industries.
Behavior change is crucial in the industry, which can only be achieved with appropriate understanding, motivations, education and a common shared approach.
Sources: pixalate.com, interceptd.com, botlab.io, wfanet.org, zorbasmedia.com