Mobile ad fraud is a real and growing problem. Since 2011, we have been provided the most comprehensive fraud solutions in the market to the world’s largest advertisers. AppsFlyer delivers a multi-layered, two-tiered solution, providing both proactive prevention against known fraud and deep insights to detect likely fraud.
As the Attribution Authority, we are the only provider to deliver cutting-edge, comprehensive solutions to today’s fraud challenges with total transparency and no conflicts of interest. At AppsFlyer, we have not ever and will never offer paid products or fraud services for ad networks or DMPs. Any provider that sells your data as a DMP/Collective, offers paid partnership programs or paid services to ad or affiliate networks has an inherent conflict of interest. This conflict of interest and bias is dangerous, undermining the trust and credibility needed to make important attribution and measurement decisions.
Our robust and constantly evolving set of capabilities aim to actively combat every type of mobile fraud at every stage, both through real-time fraud protection and deep fraud analysis.
While the responsibility and the liability for selling fraudulent traffic lies with the ad networks selling this traffic, mobile marketers often lack the expertise or resources to actively identify and address mobile fraud. The Active Fraud Suite with Active Fraud Insights provides mobile marketers with best-in-class real-time protection and fraud analysis capabilities, delivering confidence they need to take decisive action.
With that said, automated protection technologies are not bulletproof and savvy fraudsters have found ways to avoid even the most advanced protection. Our Active Fraud Insights dashboard provides marketers with the most comprehensive and useful anti-fraud data in the market, delivering the insights they need to take action.
All of our anti-fraud solutions are built and optimized on industry’s largest mobile marketing database. Proprietary big-data algorithms combined with deep industry experience and real-time machine learning and artificial intelligence combine to deliver the tools and capabilities mobile marketers need to actively stop mobile fraud in its tracks, and detect where likely fraud is hiding. With full performance data integrity, mobile app marketers working with AppsFlyer enjoy more accurate performance data and a distinct competitive advantage.
Please note that due to the growing complexity and sophistication of mobile fraud, we cannot disclose the full scope of our anti-fraud activities or capabilities. With that being said, below are some of the measures we take to prevent every type of mobile fraud.
The Vault is AppsFlyer’s proprietary, first-party anti-fraud database and the foundation of our Active Fraud Suite. This is the largest mobile app database in the world, over ten times larger than that of our closest competitor.
The Vault contains device-level metadata for over 98% of the world’s smartphones and attribution and engagement metadata for over 20 billion app installs. In total, this adds up to nine billion dollars of yearly mobile performance data measured and cataloged.
This enormous anti-fraud database powers advanced algorithms at unparalleled scale. Furthermore, machine learning and artificial intelligence capabilities monitor hundreds of billions of interactions each month, detecting new sources of fraud and types of fraud with unparalleled speed and efficiency.
With over 1 trillion events tracked every year and device-level data on over 98% of the world’s smartphones, The Vault’s cross-publisher live database dynamically identifies and blocks fraudulent IPs, user agents and device IDs. Our algorithms actively monitor billions of mobile ad interactions every day, automatically verifying legitimate activity and cataloging suspect or mismatched IP addresses, user agents and device IDs in real-time.
Unlike many others in this space, AppsFlyer uses both third-party licensed data and a massive first-party, proprietary set blacklists and filters. This is why many marketers who move their business to AppsFlyer discover significant savings and performance gains due to fraud protection in their first few weeks on the platform. In 2017, two of the world’s largest gaming publishers discovered over a million dollars of fraud in their first month after moving their attribution and fraud protection to AppsFlyer.
Our robust statistical anomaly detection, machine learning and artificial intelligence continuously learn and adapt, improving our ability to detect outliers (abnormal trends or events) and potential fraud-driven anomalies. For example, in one case we identified a suspicious set of installs. A quick analysis revealed that these were not new installs, but bot installs and app updates. This saved the client $300,000 in cost-per-install media over just one weekend!
What started as click-to-install distribution modeling (analyzing the average click-to-install time per channel, device and source) has evolved into a robust set of big-data models. Typical anomalies our system aims to detect include mean-time-to-install (MTTI), click-to-time-install (CTTI) geographic distribution, click volume by IP address and device ID, user agent versus IP benchmarks and more.
As the machine learning and artificial intelligence solutions detect new anomalies, our data scientists run regression analyses to determine the accuracy and scope of each new pattern. Once validated, these new models are added to the entire AppsFlyer database, providing the most comprehensive anti-fraud models in the world.
For example, in mid 2017, our system flagged a particular install and engagement pattern that evaded typical MTTI and CTIT protection. However, the uniformity of install and engagement times across clients and regions from a small subset of SiteIDs (sub-publishers) indicated that this was clearly fraud. In less than a day, this particular engagement pattern was flagged and blocked moving forward. This protection is only possible with AppsFlyer’s scale, machine learning and artificial intelligence capabilities.
Our industry-leading full raw data reports provide marketers with total access to all of install, engagement and purchase data. These reports feature precise timestamps for every action and engagement (including uninstall data), so you can track every reported user action. These raw data reports have been used to power publisher-specific anomaly detection and fraud protection. For example, using our raw data reports an advertiser recently discovered that their previous attribution provider missed nearly one-in-four fraudulent installs. Additionally, many advertisers and partners also use their AppsFlyer raw data reports as invoices for performance payments or to settle billing and performance discrepancies.
Furthermore, raw data reports on blocked and suspicious activities are available as part of our Active Fraud Insights offering.
Install fraud is a real and growing threat. Once activated by the advertiser, app install validation works in tandem with DeviceRank™ to prevent app install fraud on iOS. With every app installation reported by the app, we query the originating app store’s servers to validate that the install data matches our records.
As marketers shift their focus to ROI and bottom-line performance, scammers are increasingly looking to capitalize on purchase fraud. Savvy fraudsters simulate false purchase events to inflate the perceived value of their installs. As an attribution and analytics platform, we work hard to verify that every purchase event is legitimate. Our dynamic receipt validation actively detects and filters out non-validated purchase events across both iOS and Android.
Our native, secure, post-compilation SDK uses secure verification mechanisms (e.g. hashing) as well as DeviceRank™ ratings to authenticate genuine device installations and in-app events. These mechanisms are also utilized to prevent spoofed events due to man-in-the-middle intrusions and retransmission fraud.
In contrast, other SDK solutions are often easy to imitate, manipulate and simulate, opening the doors to fraudulent install, engagement and purchase event reporting. In comparison, Open Source SDKs are easily manipulated. In our internal testing, white-hat hackers were able to manipulate a popular Open Source Attribution SDK in less than 30 minutes, sending fraudulent SDK triggers and manipulating their attribution and measurement solution with ease.
DeviceRankTM is the world’s first and only device-based fraud protection technology. By rating devices based on a rich, cross-app install and engagement metadata profile, DeviceRankTM identifies fraudulent devices 3x to 12x better than industry standard fraud solutions. Installs and engagement from devices with a DeviceRankTM of C (i.e. fraudulent devices) are not attributed.
Active Fraud Insights provides the world’s most advanced marketers with unparalleled insight into their fraudulent and suspicious activity.
Put simply, this is the world’s most advanced fraud detection solution.
The Active Fraud Insights Dashboard provides an interactive dashboard where marketers can easily discover which media sources are driving fraudulent, suspicious or clean installs, perpetrating install hijacking, click flooding, DeviceID reset fraud and more. Easily determine which campaigns, GEOs, networks and SIteIDs most affected, so you can take timely, decisive action.
Raw Data Reports provide full reporting on all blocked clicks, installs and in-app events. Ideal for offline analysis and dispute resolution.
Expert Consultation with one of our dedicated fraud analysts is also available for large advertisers. Take advantage of our deep experience, proprietary internal tools and deep industry relationships to solve your toughest fraud challenges.
At AppsFlyer, we sit at the nexus of the mobile app industry, maintaining an open ecosystem that includes over 2,500 partners and 12,000 advertisers. As the industry’s largest trusted third party, we are relied upon as the Attribution Authority.
We take the trust that marketers place in our services very seriously. Though we rarely directly address others in our space, there is a cancer that is growing in this industry and that advertisers must be aware of. In an effort to boost revenues, many attribution providers are now offering paid products to affiliate and ad networks. In one recent example, an attribution provider is both selling their customer data as a DMP/Collective AND selling anti-fraud solutions and reports to networks. This is a clear conflict of interest for a number of reasons:
It should be noted that companies playing this “game” with their advertiser data and revenue models often lack the oversite or due diligence of an experienced board of investors. This is dangerous territory and something all marketers should be wary of when choosing a measurement partner.
As the Attribution Authority, we are obligated to take action when bad actors are identified. In the last year alone, we sent dozens of warnings to partners exhibiting suspicious or fraudulent behavior. Ultimately, we had to remove several networks from our integrated partner program for sending fraudulent traffic and failing to comply with our requirements. These actions are both unfortunate and imperative to the continued success of everyone in the mobile ecosystem.
The world’s largest advertisers trust AppsFlyer with their attribution and marketing analytics. Heavily security conscious companies, such as McAfee, Credit Karma and DocuSign all rely on AppsFlyer to provide the highest quality mobile attribution and marketing analytics data.
We are deeply committed to providing mobile marketers with absolute data clarity and integrity. DeviceRank™ as well as all of our core mobile fraud protection and prevention services are provided as part of our standard offering – at no additional fee. Advanced Active Fraud Insights deliver unparalled depth and actionable anti-fraud insights to those who want to detect the advanced fraudsters hiding behind well-developed solutions.
Furthermore, our pay-once per attributed install pricing provides marketers with predictable attribution and analytics costs, removing concerns of increased fees due to fraudulent clicks and in-app engagement.