This week, Stripe released The State of Online Fraud, a comprehensive review “of global fraud trends [they] uncovered after analyzing billions of attempted payments and surveying more than 2,500 business leaders.”
It’s a fascinating peek behind the curtain into the scale and sophistication of fraudulent behavior in the payments space, and sheds some light on how this criminal enterprise is a global phenomenon with distinctly regional characteristics.
As one of its authors notes, because fighting fraud is an adversarial game, they are usually pretty tight-lipped about their observations, but this report afforded them an opportunity to publish a few of their findings.
One finding in particular struck as analogous to our own work in music streaming. Stripe notes that:
“Businesses around the world experienced increases in fraud; however, businesses in Latin America were—and continue to be—particularly susceptible to fraud attacks. We observed that businesses in Latin America had a 97% higher fraud rate compared to those in North America and a 222% higher fraud rate than businesses in the Asia-Pacific region. This is due to a variety of region-specific factors, such as a locally run payments infrastructure.”
In music, you may have recently read about the coordinated takedowns of infringing services operating in Brazil by the IFPI, Brazilian law enforcement, and Pro-Música Brasil. Or of another IFPI effort from 2020 along similar lines to combat manipulation in Germany. Every market faces this challenge head on, but not equally, and that’s worth acknowledging as services operate across borders and continents.
The same Stripe author observed that, “one of the nice bits about being a multinational player in high-stakes games is that you can watch as other multinational players in the same game try to region-shop for advantage. The pace of iteration by adversaries is very intense.” We wholly agree. Analyzing streaming fraud through a window that offers us a global view, we’ve noticed region-shopping – and the creativity of highly iterative adversaries – firsthand. To be specific, we’ve seen 640% to 1000%+ higher fraud rates above the US as a baseline in certain markets. And no, those percentages are not typos.
Region Agnostic Observations
Another parallel to the Stripe analysis is that more than 75% of B2C subscription businesses – precisely what music streaming services offer – reported that their manual review load increased and that they had to divert more resources to fight fraud in the last year.
Consumer-facing businesses have more brand awareness and as a result, fraudulent actors are more likely to target them. You would be hard-pressed to find a higher profile subscription business operating globally than music streaming.
As a final note, when it comes to considering how to act, and why, the cost of fraud is often partially hidden. It’s important to recognize that “many businesses have to grow their fraud team or divert product or engineering resources to manage operational overhead, shifting valuable resources away from their core product.” Platforms all building the same solutions on their own datasets is a perfect example of undifferentiated heavy lifting, to borrow a Bezos term, that can be more effectively, and cost efficiently, outsourced to someone working across multiple datasets and building dedicated infrastructure for the task at hand.
One adaptation that is likely to emerge in streaming, and one we are on the forefront of at Beatdapp, is collaborating to build threat detection models against richer sources of data that will help businesses make faster, more accurate decisions.
There are ways to collaborate through an independent third party that allow for fierce competition between platforms to persist – as it should – while also confronting shared adversaries collectively by employing strategies, tactics, and tools that advantage the entire industry, proportionately, and as a whole.
