FraudScore Turns 10: Our Journey to a Global Anti-Fraud Platform

Ten years is a long time in technology, and it’s long enough to watch an industry change, reinvent itself, and learn a few lessons along the way. When we started working on FraudScore in 2015, ad fraud wasn't a mysterious topic. Marketers already knew it existed; many had felt its impact firsthand. But much of the market was still operating on assumptions and surface-level checks. Fraud was something they tried to “control”, not fully understand. We came from inside performance marketing. We dealt with questionable traffic, unexpected spikes, suspicious installs – the familiar signs for anyone who worked with performance traffic. We built a tool simply to understand what was happening in our own campaigns, and very quickly realized that if we wanted true objectivity, we had to make a choice. So we closed our CPA network and committed fully to building an independent product. That decision defined everything that followed.

Early years: clarity builds trust

Those first years were about proving not every spike was fraud; not every clean dataset meant safety. The industry needed clearer diagnostics, more transparency, and a shared language about what “quality” really meant. We began analyzing behavior: post-install events, patterns, clusters, the small details that reveal big things. Marketers who had estimated “maybe 5% questionable traffic” suddenly had insight showing 20%+, and in some markets much higher. By 2018, the numbers spoke loudly: on average, more than 22% of traffic worldwide showed signs of fraud, reaching up to ~40% in certain regions. It wasn’t about creating panic, it was simply about showing reality.

Growing with the market

As digital advertising matured, so did fraud. It moved into attribution paths, retention curves, and user behavior. We rebuilt our architecture, expanded to Europe, and integrated machine learning. The product became not just a detector, but a system for decision-making and fairness for advertisers, agencies, and platforms who wanted to rely on data instead of assumptions. When the pandemic shifted the world online, the challenge accelerated. In some campaigns, fraudulent activity approached 90%. Weekly manual checks were no longer enough. Real-time protection went from idea to necessity, and we built it. Later, deeper ecosystem partnerships became natural. In 2024, our integration with AppMetrica marked a new phase: fraud-risk intelligence embedded directly into analytics workflows, not sitting on the sidelines. Anti-fraud was no longer a “tool.” It became part of infrastructure.

2025 and beyond

Fraud rarely looks like fraud today. It optimized. Sometimes it is indistinguishable from the actual user behavior. In mid-2025, global fraud levels reached their highest point in five years – nearly half of traffic in some segments. So we built tools for nuance for the kinds of fraud you don’t catch by staring at a dashboard, but by understanding intent and context. Our mission didn’t change. Our methods evolved.

Looking forward

If the last decade in advertising was about scale and automation, the next one will be about trust, signal quality, and context. Fraud evolves. So do we. Grounded in data, independent in decisions, and always exploring new fraud patterns before they hit the market. We don’t think anti-fraud is about fear. We think it’s about clarity for the industry we work in and believe in. Ten years ago, we built a small tool to answer our own questions. Today, we support advertisers and partners around the world still guided by the same idea: know what’s real, and make decisions with confidence. Thank you for being part of this journey. We’re just getting started!