Crifbürgel offers a modular solution against online fraud

  • 97 percent of online shops affected: Fraud and attempted fraud are everyday business for many online retailers in Germany.
  • Exposes bot networks and scammers: AI-supported plug-in takes over identity and fraud checks and thus protects against misuse.
  • Reduced manual individual processing: Machine learning reduces clarification cases compared to rule-based systems by over 80 percent.

One in three Germans orders online at least once a week. This carries risks because credit card fraud, identity theft and account takeover are the rule rather than the exception on the Internet. In Germany and Switzerland, 97 percent of online shops have been victims of fraud before. This is shown by the "Fraud in online trade" survey by Crifbürgel in spring 2020, for which 110 online and mail order companies in Germany and Switzerland were asked, among other things, about possibilities of fraud prevention.

According to a study by LexisNexis, there were 277 million online attacks worldwide in the first six months of 2019 alone, an increase of 13 percent. However, these are only the human-initiated attacks: In the same period, attacks by so-called bot attacks against online retailers increased by 305 percent. Fraudsters use accounts created by bots, for example to test online identities or credit card data and to redeem virtual vouchers, whereby their approach is becoming increasingly progressive and more difficult to identify.

Fraud prevention: Most often, customer name or address data is falsified

Fraudsters most often use falsified name or address data, 71 percent of shops know this. Then identity theft follows: 68 percent of retailers have already experienced that a customer claims to be a completely different, real person. 42 percent were affected by friendly fraud. This is the name of an order in which the buyer knows in advance that he cannot or does not want to pay the invoice, but still receives the goods. At 52 percent, more than half of fraud cases are the result of stolen payment data, such as credit cards. Millions of dollars of damage occur every year.

“Fraud prevention is therefore becoming increasingly important because data theft not only concerns online shops, but also providers of digital services such as banks, telecommunications and mobility providers as well as hosting and web services, streaming services, gaming and comparable providers – wherever fraudsters can misuse information to gain economic advantages, ”explains Kai Wanka, Head of Fraud Solutions at the information service provider Crifbürgel.

Crif Bürgel
Kai Wanka is Head of Fraud Solutions at Crifbürgel. (Image: Crif Bürgel)

Answer the right questions with artificial intelligence

Protection for providers and their customers is provided by Hybright, the new software solution for risk assessment and fraud prevention by Crifbürgel. The machine learning-based check of customer behavior, identity, location and device data detects any abnormalities during an application or order process and triggers an alarm. “With every interaction on the web, the digital footprint of every customer grows. And our tool tracks it like a motion detector. ”Using behavioral biometrics, predictive analytics and automated pattern recognition, good and less good customers are differentiated more clearly and fraudsters are screened out before they come into play.

"For a top customer experience and good fraud prevention, providers need to know: who is my customer, what does he want and how does he behave – typically?" Explains Wanka. “But no online provider can know all of its customers as well as a corner shop in its corner shop. Even if he has very, very well-maintained customer relationships – how can he suspect that a fraudster has taken over the account of a good customer? And who, let's say, has 100,000 users in the online shop, cannot always have all the information ready to effectively prevent fraud, ”continues Kai Wanka.

Fraud prevention: Identifying misuse through the identification and verification process

Hybright can immediately identify misuse with a coordinated identification and testing process and take appropriate countermeasures. The results are displayed in an individually configurable online cockpit, which presents all KPIs relevant to the user in a dashboard. All transaction-relevant information is clearly summarized in a numerical score and, if desired, in a traffic light scale.

"These are clear recommendations for action. No provider has to forego the popular payment on account from the outset. Instead, they can use transparent information to decide how they want to deal with suspected cases and reinsure themselves with the customer via the order, offer certain payment options or completely reject a transaction, ”says Wanka.

With the intelligent Fraud Prevention Assistant, online providers always have full control: "Monitoring transactions and fine-tuning their individual rules remain simple and intuitive, while providers take full advantage of complex and evolving processes."

Fraud prevention: Reduce clarification cases with AI

With the help of machine learning, Hybright continues to develop according to the requirements of the user. In test operation, the clarification cases were reduced by over 80 percent compared to rule-based systems. "With this technology, online providers can observe patterns in customer behavior, get to know their customers and better meet their needs, and be a big step ahead of fraudsters," explains Wanka.

Fraud prevention

To distinguish fraudsters and trustworthy customers in real time, Hybright also uses ThreatMetrix's digital identity information, giving it access to a global data universe and the information behind 150 million authentication and trust decisions every day. For the best possible protection, the test is carried out in parallel on five levels, over 900 different factors are tested:

  • Identity proofing: Is there something wrong with the user? Is he real, has he ever been a customer and what about his credit rating?
  • Customer Order History: Is something wrong with the purchase? Do the frequency and value of the goods correspond to the typical behavior, is an unusually large and expensive purchase or paid differently?
  • Device Intelligence: Is there something wrong with the device? Are there any irregularities in the IP address, browser, location or in the typing and clicking behavior?
  • Account Protection: Is something wrong with the customer account? Does the behavior match the existing data or has master data, for example email address or home address, been changed?
  • Transaction execution: Is something wrong with completion and delivery? Do the payment details and address match previous purchases and other sources, or may the shipment be redirected?

Hybright is part of the newly created growth engine from Crif Bürgel. With this ecosystem kit, companies from various industries can jointly leverage cross-selling potential and offer their customers a broader range of products: the as-a-Service platform connects providers and their offers, recognizes cross-selling events and handles data exchange between the partner companies ab – in accordance with all legal requirements and high security standards. (sg)

Also read: Online fraud: How online retailers can protect themselves with AI