Merchants have the option of disputing chargebacks they believe were filed deceitfully, in a process known as Chargeback Representment. The process requires gathering a wide array of data points at the time an order is made. In the case of a fraudulent chargeback, the relevant points that can prove an order was indeed made by the cardholder who filed the chargeback are submitted in the form of Compelling Evidence. The high burden of proof renders the process both resource intensive and expensive. However, when streamlined and embedded in the business flow, it can help businesses recoup lost revenue. To become better acquainted with the types of information that can help differentiate between unauthorized and authorized transactions, and used as evidence to effectively dispute chargeback claims, we have created this comprehensive guide.  

Excessive chargeback programs

Merchants who surpass a chargeback rate threshold set by credit card issuers are penalized by enrollment into an Excessive Chargeback/Risk Program. The terms of these programs vary between issuers, and depend on the degree and persistence of high chargeback rates, but most penalties include some combination of fines, higher processing fees, and mandatory risk education programs. Merchants can get actionable advice on preventing chargebacks here.

Chargeback guarantee

The chargeback guarantee is a business model under which fraud management solutions assume liability for orders they approve. If an approved order turns out to be fraudulent, merchants are reimbursed for the entire chargeback amount within 48 hours. Different solutions cover different chargeback codes, and it’s important for merchants to understand which scenarios are covered and which aren’t.

Machine learning

Machine Learning is an advanced artificial intelligence technique that allows computers to refine their behavior without being explicitly programmed. There are many subfields of machine learning, but behind all of them is a basic idea: rather than telling a computer how to solve a problem, show the computer relevant information and let it figure out how best to solve it.
We show a computer millions of orders – all of which are tagged as either legitimate or fraudulent – and ask it to determine retroactively how order data points could have best been considered and weighted to arrive at the correct assessment. In either case, the resulting algorithm will not be developed by rules, but rather based on trends across millions and millions of online shopping orders.

Click here to learn more about machine learning and how Riskified uses this technology to facilitate accurate fraud detection.