Moshe Vaknin, co-founder and CEO of growth marketing platform YouAppi, looks at the growing problem of mobile ad fraud, and what measures can be taken to prevent it.
The last few years have seen a revolution in the business of app marketing. With over a billion apps in each of the two major app stores, app developers have had to get smarter in how they market themselves. And it’s not just about driving downloads; it’s about driving downloads from the types of people who are likely to stay loyal to the app and generate revenue for the developer.
We’ve built our business on this revolution. Our OneRun platform, which has been four years in the making, combines the power of machine learning with our proprietary predictive algorithms and cohort technology to analyse the mobile content consumption patterns of over 2bn users, converting data into profitable customers. And it’s gained us real traction in the market. Collectively, our 450 advertiser clients have run more than 15,000 mobile user acquisition campaigns on our platform, and they come back to us because it works.
A dark cloud
But there is a dark cloud on the horizon that impacts everyone in the digital advertising business, and the name of that cloud is fraud. According to figures from the World Federation of Advertisers, the cost of ad fraud will rise from current levels of $7.2bn (£5.5bn) to at least $50bn by 2025. That’s a lot of money going into the pockets of bad actors. And while that figure includes both desktop and mobile ad fraud, the rapid growth – and relative immaturity – of mobile advertising, make it a soft and particularly appealing target for the fraudsters.
It gets even worse when you consider the knock-on effects of this activity, primarily the rise of ad blocking. There’s a direct correlation between the rise of ad fraud and malware, and the rise in ad blocking, as consumers, understandably, seek to protect themselves. A January 2016 study by PageFair found that 419m people worldwide use mobile ad blockers. That’s a fifth of the world’s smartphone users (and represents an increase of 95 per cent from the previous year). If ad blocking, driven by fraud, continues to rise, the fundamental economic model of the web will be undermined, and the day ad blocking comes in earnest to apps … well, the implications are obvious.
Fighting back
It’s because of this that 12 months ago we took the decision to do something to fight mobile ad fraud on our OneRun growth marketing platform. Fraud is a manyheaded monster, and while there’s no silver bullet, our data scientists are hard at work identifying the many different ways in which mobile ad fraud is perpetrated. They then come up with a solution to defeat the fraudsters. We’re pleased to say that we have put in place a solution that will enable fraud prevention and protection for all our clients.
Our fraud offering currently consists of six different diagnostics tests, including two that our data scientists think are especially important: multiple-user click analysis and click-to-install delay analysis. These two are critical to our fraud prevention offering because they monitor user behaviours to identify fraudulent activities.
Multiple-user click analysis is a diagnostic test that involves our data scientists monitoring user click patterns. They look for multiple users on the same IP address and multiple clicks on ads or elsewhere within a short period of time, to detect signs of malicious downloads. More than 40 per cent of the click traffic of some fraudulent publishers is actually multiple clicks from a single user to multiple campaigns within a few seconds. Our tech can analyse and identify this behaviour to avoid wasted ad spend.
In a ‘clean’ campaign, over 90 per cent of installs occur within a click-to-install delay of four hours or less. Fraudulent campaigns look very different, with a much more consistent pattern of installs and just small spikes of click-to-installs within the first two hours. YouAppi’s data scientists use click-to-install delay analysis to identify when fraudulent mobile downloads are taking place, and put a stop to it.
Cat and mouse
With these measures in place, we feel our clients are as protected from fraud as they can possibly be. When fraudulent activity is detected in real-time, we blacklist the publisher immediately, but then work with them to identify and eliminate the source of the fraud so that they can start earning again. We appreciate that they don’t deliberately invite fraud, but as they look for new sources of traffic, it inevitably happens. What’s important is to eradicate it as quickly as possible and get them back on track.
Though we’re encouraged by the testing we’ve done, we at YouAppi understand that the fraud-prevention solution we’re rolling out now only addresses the relevant issues in mobile fraud today. In six months, we’ll be adding new fraud-fighting functionality to address changing dynamics and new technologies. Fraud is criminal activity and the criminals are already searching for the next way to ‘game the system’. As solutions to current fraudulent activities are implemented, scammers search for new ways to deceive. It’s a cat and mouse game, but by being diligent, I believe we can greatly reduce fraud.
Beyond our efforts to fight mobile fraud, the only way to truly solve this industry-wide problem is for the entire mobile marketing ecosystem to work together. Many players in the industry benefit from fraudulent clicks – not just the fraudsters. Therefore, to truly combat the issue, we need to work together as an industry in order to collectively monitor and rein in fraudulent traffic and other forms of mobile fraud.
We recognise that the fight against ad fraud is a journey, not a destination. Whatever measures we put in place, the fraudsters will try to find a way around them. But we won’t be complacent in the fight against fraud and we won’t stand still. The very future of our business, and indeed our industry, simply won’t allow us to.
This sponsored article first appeared in the September 2016 print edition of Mobile Marketing. You can read the whole issue here.
