Zensed is an anti-fraud company based in Europe with representation in Sweden, U.K. & the U.S. Its founders operate high risk payment gateways around the globe and originally developed Zensed technology in 2015 to combat their own fraud problems. Victor Fredung is the Sweden-based CEO and co-founder of Zensed. He was recently interviewed by about-fraud.com to provide some insight into how the Jell-O-like versatility of Zensed’s machine learning fraud solution differs from those of its competitors.
Starting with High-Risk Payment Gateways
RS: Why did you create Zensed?
VF: I and the other co-founders had much experience in managing high risk payment gateways across the globe for online gambling, digital goods, vouchers, etc. Operating these high risk multi-client gateways provides us with great insight into the ever-evolving world of fraud.
Over the years we have tried many new and promising anti-fraud tools in an attempt to fight fraudulent activity. However, the challenge of keeping within a 1% chargeback threshold was a constant battle impacting our pre-paid and gift card business in many ways. Short-term improvements were soon overtaken by increased fraudster sophistication.
A different approach was needed so with we took things into their own hands and developed the Zensed solution. Our anti-fraud results on our own sites were so good that we implemented Zensed for allied companies in 2016 . The results for our allied companies were equally as good and their fraudulent transactions dropped by 50% or more. Zensed was launched to the wider commercial market in 2017.
During the two-year build phase the system went through major validation and testing with our online gambling and gift card clients, helping refine the platform features. Zensed handled millions of transactions and pieces of interaction data during this period that aided the machine learning’s continual development.
Zensed today is a truly automated, real-time fraud prevention system without any reliance upon a human analyst. The platform can detect anomalies and patterns in fraudulent behavior to predict bad transactions before they go through. The name Zensed comes from the combination of the words “Zen” and “sensed,” which conveys the enlightened perception of fraudulent transactions provided by this artificial intelligence solution.
RS: What industry verticals have you worked with?
VF: Our early successes were with clients in high-risk markets such as online gambling, digital goods, vouchers and a freelancer marketplace. However, Zensed can fight fraud for any client and can tackle fraud types besides card-not-present fraud, such as promo abuse and account takeover (ATO) fraud.
RS: What data do you incorporate into your artificial intelligence solution? How does this differ from other providers?
VF: There can be different fraud signals and data points that can give an indication of what might be a fraudulent transaction. Individually these fraud signals don’t tell us that much about a transaction but combined with hundreds of different pieces of data and together with advanced analysis and custom-built modeling Zensed can provide a really accurate result.
The problem with most traditional fraud prevention tools is that they actually block too many legitimate customers because of using too few parameters when a transaction occurs. With Zensed our “false positives” are close to zero.
To give an example, if a customer makes a purchase from an IP address that is not close to their shipping address most traditional fraud prevention tools would decline that purchase. For Zensed we use hundreds of different fraud indicator signals so an anomaly in IP address compared to their shipping address wouldn’t automatically impact the prediction as a decline. Such a case would just change the risk score calculation in the Zensed modeling for that transaction that can then be further altered by results from other data point factors.
Rigid Rules vs. Jell-O-Like Adaptability
RS: Some of the vendors on the market provide black-box off-the-shelf solutions. How does the compare with your approach to machine learning?
VF: The main difference between traditional fraud prevention tools and how Zensed prevents fraudulent transactions is the much larger number of factors and data points we take into consideration before predicting the outcome of a transaction.
Traditional fraud prevention tools are opaque and secretive as to how they serve client results as they usually only work with a few different factors such as IP address, white-listed and black-listed IP addresses and how many purchases has been made from specific customers. This is typically called a “rules-based” fraud prevention engine where the operators of the payment processing decide on specific rigid rules to allow or deny purchases.
Most leading anti-fraud solutions are too rigid and have a one size fits all approach to clients. Zensed’s experience is that all clients are different and our technology and AI models treat each customer differently. Instead of a box that is one size fits all, Zensed is more like Jell-O. If your fraud falls outside our initial model don’t worry, we will mold the Jell-O to catch it.
On a feature level, there are also several things that distinguish Zensed from the competition:
No JavaScript – All of Zensed´s major competitors require the client to be integrated with a JavaScript plug-in to track and monitor customer behavior patterns on the checkout page. Zensed does not.
Custom Data Fields – We are the only company we know of that welcomes adding additional information to our standard data fields. Our competitors typically restrict the data they accept to a set number of fields. We want more information because the more data we receive the better our artificial intelligence engine will be.
Flexible Integration – Zensed is an out of the box service which offers a wide variety of integration options through our RESTful API. We don’t want to turn away clients and can therefore modify the system to fit specific customer needs.
Support – It’s not a feature of the software, but it’s important to highlight that we are online 24/7 to fulfill customers’ needs. If you have a technical query we are here to help you fix that.
RS: What’s your pricing model?
VF: Zensed does not rely on third party data providers for our core service meaning we can provide Zensed technology at competitive rates. Our commercial terms are priced on a per client basis with the expected volume of transactions a major consideration. In general, the cost of running the Zensed solution starts at a few cents per API query with discounts depending on the volume of usage.
Reducing Fraud Rates by 50%+
RS: How effective is Zensed in reducing actual fraud and how effective is it reducing false positives (i.e. the insult rate)?
VF: Zensed was initially benchmarked against two industry leading fraud detection solutions that identified approximately 20% of fraudulent transactions that would have resulted in chargebacks. Using the same data, Zensed found 60% of the fraudulent transactions on the first use of the technology. Shortly afterwards as the machine learning evolved the anti-fraud modeling, this increased to 80% of fraudulent transactions detected.
Each of our clients have benefitted from at least a 50% improvement in reducing their fraud rate when they started using Zensed. Over time, with the continual evolution of the Zensed models used by customers their fraud rate reduction improved between 65% and 80% from before Zensed.
Important as well, Zensed reduces false positives to negligible levels which helps maintain client brand reputation and positive customer relations by not turning away good business.
RS: There are quite a few solutions that claim to incorporate machine learning. What do you think is the future of fraud prevention technology beyond generic machine learning?
VF: Our belief is that the rigid anti-fraud technologies are a thing of the past. Fraud prevention has to be client-led as each customer is different. A new generation of flexible, service-focused technology companies harnessing big data will provide new anti-fraud technology to help clients stay one step ahead of the ever more sophisticated fraudsters.
RS: What type of clients would want to use your solution instead of some of your better known competitors?
VF: Zensed can be used by every type of business to address their fraud problems, not just financial services providers like, banks and acquirers but even for loan providers and law enforcement.
We have already had customers approach us who had been using our larger competitors but felt they lacked support and guidance from them. We can honestly say that we have never lost a client because of the results we provide.
