Mobile fraud is big business: here’s how game developers are fighting back

Mobile games can easily attract millions of downloads. At the same time, there are a lot of mobile game developers out there, so it’s unsurprising that developers have turned to advertisers to help them market their product. Gaming advertisers often have significant marketing budgets at their disposal, making the mobile gaming industry particularly interesting and attractive to fraudsters.

The Mobile Games Fraud Problem

Goodgames’ first mobile title, Empire: Four Kingdoms, became one of the most popular games in the strategy genre; Its newest titles, Big Farm: Mobile Harvest and Big Company: Skytopia, were installed by millions of users worldwide. In order to acquire new users, the developer uses a combination of Facebook, Google Ads, and preload campaigns (where the app comes preloaded on the mobile phone) to ensure downloads and retention. In addition to that, the company relies on various ad networks and agencies to keep the best inbound funnels alive while killing the rest. For that reason, as you can imagine, fraud is an incredibly important thing to be mindful of.

Goodgame Studios has been dealing with fraud threats for several years, but the threats keep getting more complex, according to Tim Heicks, Goodgame’s Head of Performance Marketing. “Although our internal fraud scripts already detect various fraud types, these days we are noticing that newer, smarter techniques are popping up,” says Heicks. “Fraudsters have been using this as an opportunity to mix different types of fraudulent installs together by blurring the lines and making it harder to identify.”

In many ways this leads to an ongoing battle between the fraudsters and the advertisers — a conflict where the advertiser is at a disadvantage since fraud detection isn’t exactly why they got into the mobile game business to begin with. That’s part of why there’s been a growing industry around fraud detection: developers and publishers need to surround themselves with strong anti-fraud solutions.

The Evolving Fraud Landscape

In 2015–2016, the fraud problem was already notable, but the industry was dealing with mostly primitive types of fraud, such as click-spamming and click-injections, according to Inna Ushakova, CEO & Co-founder of Scalarr. While Ushakova says these types of fraud are still common, “smart fraud” is the newest threat were smart bots or sophisticated bots can fully emulate user behavior and even make in-app payments. Ushakova also considers modified click-spamming as a form of smart fraud. “This year, we have seen significant growth in mixed traffic, which is one of the most insidious types of fraud these days,” she explains. “You might be facing click-spamming mixed together with bot traffic. From the developers’ point of view, it may look unsuspicious because of click-spamming, but in fact, this is a prime example of disguised fraud.” Ultimately, the biggest problem with fraud these days is that there’s so much money in fraud that perpetrators are incentivized to find new ways to fraud game developers as old mechanisms stop working.

With the evolving threat of new fraud types, Goodgame Studios’ scripts and anti-fraud tools were only catching primitive types of fraud. The developer then decided to survey the market for more advanced tools and came across anti-fraud solutions based on machine learning algorithms, leading their discovery of Scalarr.

Scalarr uses machine learning and big data algorithms to fight app install ad fraud primarily because of their ability to analyze a huge number of metrics and interrelations between them. This allows Scalarr to make more efficient analysis than manual human or rule-based techniques, which are still an integral part of most solutions. The end result? By reducing the number of false-negatives and false-positives, Scalarr’s accuracy is currently up to 97%, according to Ushakova.

The Road Ahead

Goodgame Studios is able to discover specific percentages for each type of fraud. The company has reported that 77.7% of all fraud cases were caused by classic bots, with click injections responsible for 14.8% and classic click-spamming responsible for 7.5%. But among this 7.5% of click-spamming, about 97% can be classified as modified click-spamming. “The biggest problem behind modified click-spamming is the fact, that we were considering these fake installs to be real, and we were paying for them because of their ‘realistic’ behavior,” explains Heicks. “But now we have a better interpretation of this type of fraud and have higher confidence buying traffic.”

Despite all the hype around mobile ad fraud, the problem remains and tends to grow: Scalarr estimates the damage caused by app-install ad fraud could reach $4,6 billion in 2018 with further compound annual growth rate of 20 % from 2018 to 2020. But this negative tendency could change with more serious and uncompromising approaches to implement strong anti-fraud solutions. As long as there’s money in developing successful mobile games, there will be money in fraud. Luckily, there will also be anti-fraud solutions that rise to stem the tide.

Originally published at www.inc.com on August 16, 2018.