What is mobile ad fraud?
In an everyday battle for users a mobile advertiser is maintaining, the problem of quality traffic seems to become increasingly important.
Mobile ad fraud is a practice of cheating the advertisers by
- serving ads that have zero chance to be viewed by users. Advertisers who work on a CPM basis are at risk. CPM or cost per mile payment model means that advertiser pays per one thousand impressions.
- serving fake or low quality clicks. Advertisers who work on a CPC basis are at risk. CPC or cost-per-click payment model means that an advertiser pays for clicks the user makes upon the ads.
- serving fake or low quality installs. Advertisers who work on a CPI basis are at risk. CPI or cost per install is the payment model when advertisers pay a publisher for delivered app installs.
IAB reports the $1.3 billion cost of mobile ad fraud to advertisers worldwide in 2015. For both mobile and desktop traffic, 70% of cost relates to CPC and 30% to CPM pricing models. Mobile app advertisers are interested in installs more than in clicks. CPI is higher than CPC, and it turns the installs into a perfect target for cheaters. According to AppsFlyer’s research, this year the advertisers will lose up to $350 million because of mobile app install and engaging fraud.
Types of app install fraud
Fraud in CPI pricing model can be divided into 2 groups:
- The frauders manipulate with a post-back event code, and system believes the app was installed while it wasn’t.
- Installs are made by bots, not users.
- Installs are made by users from co-called human farms (people from developing markets are gathered, given smartphones and are paid for manual operations with installing and actions). These users have no value.
Low quality traffic
- Installs are made by humans, but they are not a targeted audience and are not actually interested in your app, or an advertiser did not express interest in this audience. Such undisclosed re-brokering happens when there are too many counterparts, and process of the buying-selling of an offer is so complicated that an advertiser doesn’t know where the install had been performed.
According to AppsFlyer, verified fraud will cost advertisers $100 million and suspected fraud will cost advertisers $250 million in 2016. Botnets (groups of bots hosted in many computers) can really damage mobile advertiser’s marketing budget. These include both financial losses and low KPI as ROI, ARPU and LTV.
How to detect an app install fraud?
There are some signals the advertiser can recognize by himself:
- Enormously high download rate from one source
- Over-short period between click and install
- Identical behaviour patterns for the group of installs (install-opening-action with the same time period or same sequence of in-app actions)
- Different geos for click and install
- Bulks: numerous installs from the same device ID or IP address
- Click IP = Advertiser’s IP
- Other inconsistencies, e.g. difference between click and install geo, difference between geo of download and app language settings, high install rate at night, etc.
These signals can be more or less easily recognized if an advertiser has 1 or 2 apps. If there are 10-20 running campaigns with 20+ publishers, the advertiser will have to monitor 200+ placements.
The further it goes, the more technological the bots become — they’re not only installing the app, but performing some actions and simulating real user’s activity. Sophisticated attacks can be identified only with help of big data and machine-learning algorithms.
How an advertising platform can prevent fraud?
Clickky serves the interests of advertisers by monitoring the delivered traffic on a daily basis. As bots are using more and more sophisticated algorithms, we are starting to use Big Data based decisions provided by Forensiq. Forensiq, a world famous anti-fraud system, will ensure Clickky’s mobile advertisers the additional real-time traffic quality control over the pre-campaign, campaign and post-campaign period.
Forensiq provides Clickky’s advertisers faster and deeper monitoring of the publisher traffic and feedback collecting. Clickky gets more power to exam new traffic sources, define both fake and low quality traffic and remove them promptly. Forensiq uses a multi-layered approach to fight fraud, and among its methodologies there are fraud intelligence database, proxy unmasking technology, device manipulation detection and other.
Clickky and Forensiq answer the fraud challenge together by prioritizing the delivery of quality mobile traffic to advertisers and maximization of their ROI.