2018 for FraudScore: perspectives, results and new directions
As we are all already in the new year, FraudScore decided to share our story for 2018 and provide a picture of what changes and trends have been present in the e-marketing industry, and particularly when it comes to fighting fraud.
2018 for the industry
According to WARC, in 2018 the budgets spent by companies on digital advertising have reached a number of almost 560 billion USD, and the forecasts say that this number shall increase at least by 13% by 2022. Juniper Research has stated that mobile internet has proven to be the fastest growing ad medium in 2018.
Additionally, 75% of marketing budgets are spent on mobile advertising, and therefore, we can all understand how attractive these funds are for fraudsters. It’s not a secret that the bigger the marketing budgets for advertising are, the more sophisticated fraudsters methods get. Latest FraudScore statistics for 2018 show that an aver. of 22% across all international ad traffic (both mobile and web) is fraudulent. If we take a deeper dive into the geo regions, we’ll see that the numbers vary between Russia and CIS with almost 40% fraudulent traffic to 17.64% in EU.
We know that these statistics might scare you, but the truth is bitter. Comparing to 2017, the growth of loses that fraudsters have caused to marketers has reached a skyrocketing number of $3.9 billion.
Along with the evolution of fraud the industry has shifted to a new level of problem solving and recognition in 2018. We suppose that 2018 can be called “A year of education” in the aspect of fighting fraud. More and more companies have recognized fraud-detection instruments as a vital part of business-relations with their partners. Just to compare - the number of Google search requests for “ad fraud monitoring”, “ad fraud detection” and ”ad fraud prevention” has grown by 5.4 times this year. Marketers have become more eager to work with networks, affiliates, and agencies who are open to fraud detection instruments and are ready to negotiate if fraud is detected.
As an example of the industrial shift towards fraud awareness - just remember the latest action that Instagram took to ban fraudsters and increase guarantees for advertisers.
Or the whole Kochava-Google-Cheetah Mobile case when even Google took actions and made a public statement for companies who are using malicious SDKs warning them to restrain from ad fraud schemes. And Google also made a statement that in 2019 they will proceed towards improving Google Play Referrer API to prevent such cases of attribution fraud.
Even governments have recognized ad fraud as a malicious activity that needs to be prevented and punished. The latest case is when the US Government charged fraudsters with more than $26 million for fraud click advertising schemes.
WFA (World Federation of Advertisers) has issued a manifesto on the improvements that should be made to the whole digital advertising industry. One of the main changes that WFA recognizes is the problem of ad-fraud and the ways to stop it to reduce the losses of marketing budgets.
FraudScore felt it too - first, our product has become a standard participant of business-agreement between our clients and third-parties. More than 47% of our customers have added FraudScore as a mandatory side tool to the business decisions and arbitrage. Transparency in fraud fight is vital for the development of digital ad industry.
Fraud Detection in 2018. FraudScore statistics.
At the end of 2017 we’ve stated that attribution fraud was the most widespread fraud type on the market. Click-spamming and click-injection were the leading fraud tactics among 300 000 000 conversions we’ve scored in 2017. But our prediction for 2018 was that smart bots shall take the leading fraud-generating positions.
We can now surely say that 2018 has been a year of botnets. Each time you check there's more and more sophisticated malware. In a matter of days, these bots are implanted in thousands of devices to exploit profitable ad campaigns amongst other malicious purposes. FraudScore statistics estimate that in 2018 the growth of PROXY fraud, one of the symptoms of bots, in mobile traffic has increased by 1.5% compared to 2017:
The proxy fraud leaders are US and CA regions. SBWIRE latest report also stated that US is the world’s top region for bots fraud. Due to skyrocketing increase in smartphones penetration all over the world, development of IoT and accelerated growth of smart bots technologies, the botnet market is anticipated to grow exponentially over 2018-2024.
FraudScore uses 100+ different metrics to detect fraudulent and malicious activities for both mobile and web traffic. To make data more simple to manage and to illustrate the most common fraud types that are detected by FraudScore, we’ve grouped the metrics into “Fraud Reasons” categories:
- Detected proxy - all violations identified by IP addresses in conjunction with other parameters of conversion: botnets IPs, adware IPs, false GEO, traffic from proxy devices or networks etc.
- Device Software Anomalies - detected abnormal device distribution: device models, browser versions, operating system, etc.
- IP distribution anomalies - violation identified by IP abnormalities: multiple conversions from the same IP or the same IP subnet
- Suspicious attribution - click spamming, cookie stuffing, click injection, etc. This includes all kinds of fraudulent activities aimed to steal organic conversions
- Datacenter IP - traffic that originates from data centers servers or known cloud platform providers
- Device abnormalities - all violation that are identified by abnormal device parameters: device emulators, fake device user agents, IDFA/Android IDs, MAC addresses etc.
- Bad source - all that identifies fraud traffic sources: suspicious web sites, source traffic differs from the declared source etc.
Speaking about fraud reasons, FraudScore statistics show that PROXY fraud indeed has been the most widespread fraud type in all conversions processed by us in 2018. Operating system fraud and IP share the second and the third place in “FraudScore fraud reasons rating for 2018”. Click spamming and click injection (attribution fraud) have decreased by a total of 2,1% of detected fraud compared to 2017, but still attribution fraud is among the four leaders:
If we dig deeper, we’ll see that every region has a specific fraud reason break down. This basically means that fraudsters use different fraudulent schemes and in different variations for different geo regions, countries and ad campaign types. We’ve decided to share the fraud reason distribution by geo regions:
To learn more about FraudScore fraud reasons - follow the link to our helpdesk or read our blog.
FraudScore 2018: Mobile traffic statistics
2018 has also been marked as the year when an average user has spent at least 32 whole days in his mobile phone. Android devices all over the world have outnumbered iPhones, the mobile ad market has scored 26.9% growth and the numbers look even more promising for marketers in 2019. But comparing mobile ad fraud statistics for 2017 and 2018 we can say that the numbers almost doubled. We’ve gathered our mobile traffic data to illustrate the scale of fraudulent activities by OS and regions:
Fraud reasons distribution for mobile traffic shows that OS and proxy fraud have been the top issues for fraud detection in 2018:
The break down of fraud reasons by countries show that proxy fraud is the most detected fraud in US, Russia, and CIS; whereas the attribution fraud is still at the TOP in India. Fraudulent IPs are detected mostly in the Middle East and OS fraud - in China:
FraudScore updates: new features and technologies
Looking at the statistics, it’s obvious that the market for fraud detection solutions is growing and demanding a rapid development. Fraudsters evolve together with technologies and marketing budgets. Tech solutions need to be ahead of cheaters and predict the fraud patterns that are just beginning to surface or might occur.
FraudScore is always on the cutting edge of state of the arts technologies. In 2018 we’ve added more features to our core products to cover the whole advertising funnel:
FraudScore.Action - In 2018 we added 15 new patterns for CPA, CPI and CPL campaign analysis and managed to widen the spectrum for fraud patterns detection.
Smart Reject for real time analysis and fraud prevention - a new popular feature that we have developed and released in 2018. By using SmartReject you’ll have FraudScore conversion analysis almost in real time. Fraud will be detected and conversions will be rejected directly in the marketing platform. The product eliminates the need to manually reject conversion marked as “fraud” – it is done automatically. Affiliates won’t see those conversions because they’ll be rejected almost instantly. SmartReject is not just a simple one-way-integration when FraudScore receives data and we analyze it on our side – it’s a way to reject fraudulent conversion right in the marketing platform.
In 2018 we’ve launched a new warehouse for post-install events that is capable to cope with high loads and any amount of events for analysis. We are already using post install activity data for install scoring and we are now in the middle of developing the report to visualize data.
We are ready to announce the launch of our new platform in Q1, 2019. It shall combine all new features that we have been working on to take FraudScore to another level of fraud detection efficiency:
- post install events analysis
- determination of types of analysed offers and unique statistics for each analysed offer
- new applications of FraudScore reports and various options for benchmarking, fraud analysis and decision making
- new fraud detection algorithms
- And much more
More to come
We consider that 2019 shall be even more intense and vivid for FraudScore - new solutions, new launches, new countries, new partnerships. We are looking forward to partner with the industry’s leaders because we know that education and transparency is essential when it comes to fighting fraud. Marketers, developers, mass media can make a real shift - working together, sharing knowledge and developing standards, we are able to develop reliable and highly efficient instruments to confront fraudsters.
We’ve asked our CEO, Dmitry Isakov to share his opinion on 2018 and what to expect from FraudScore in 2019:
“FraudScore is always on the frontlines of fighting fraud. In 2018 we’ve managed to reach all the goals that we’ve set for this year - increase the efficiency of fraud detection, broaden the abilities for prevention and develop and release SmartReject solution. There is more to come in 2019 - the launch of our new platform in the first quarter, new partnerships and service for FraudScore clients. We’ll keep fighting and cleaning up the mess fraudsters cause to marketers. You don’t mess with FraudScore!”
PS: Let’s talk numbers
FraudScore is one of the industry’s most recognized fraud detection platform