Businesses face a constant and complex battle against fraudulent transactions. Something that makes this even more complicated is the ever-present risk of offending and turning away legitimate customers, due to overzealous or poorly configured fraud prevention measures.
This takes us into the realm of false positives and the aptly named “customer insults”. The reality is that customers are insulted when they’re prevented from completing a genuine transaction. According to PWC, “32% of all customers would stop doing business with a brand they loved after one bad experience.”
It’s clear that every business must take steps to reduce customer friction and do all they can to minimize false declines. These are estimated to cost businesses over $443 billion in lost sales every year, according to Cardinal Commerce. In fact, it’s often said that businesses lose multiples more to friction and false declines than they do to fraud itself.
So, what’s the solution? How can companies strike a balance that minimizes fraud and reduces customer insults?
Perhaps surprisingly, the answer is to make intelligent use of dynamic friction – as this article explains.
What is Dynamic Friction?
Dynamic friction works on the principle that not all friction is a bad thing – especially if it’s mostly invisible to genuine paying customers. These two rules reflect the thinking behind the process:
– Friction that introduces obstacles to likely fraudsters is good.
– Friction that obstructs a genuine customer journey is bad.
Many checks and balances can contribute to dynamic friction without inconveniencing genuine, low-risk customers. For example:
– Merchants can enforce stricter checks on new customers than they do on existing customers.
– Systems can refer back to previous transaction histories to verify if a new transaction fits the recognized behavior of an established customer.
– Businesses can make use of systems such as 3D Secure to “fast track” lower-risk customers.
– Companies can use verification techniques such as geolocation, device fingerprinting, and digital footprinting to assess the risk level of each transaction.
A combination of all the data gathered and checked throughout the customer journey can be used to create a fraud score as explained by SEON.
This, in turn, can help determine whether a transaction is allowed to move through the system with no (noticeable) friction. Alternatively, if flagged as high risk, it can be rejected outright or flagged for additional security checks or manual reviews.
What is a Fraud Score?
A fraud score is produced from all of the available information about a customer and a transaction.
For instance, the following are examples of factors that could result in a higher fraud score, representing a greater risk of fraud:
– The customer is logged on using a VPN.
– The email address provided by the customer doesn’t fit the profile of an address in regular use (for example, it may not be linked to a typical number of social media and other online accounts).
– The transaction value is suspiciously high, or is for product(s) typically targeted by fraudsters.
– The customer’s IP address is on a spam blacklist.
– The customer’s geolocation doesn’t match their stated location.
Businesses can freely decide which indicators should most impact a fraud score, and their own risk threshold levels. Once these are passed, a transaction can be blocked or flagged for manual review. For example, a score between 20 and 30 could trigger further checks, but a score above 30 could result in an immediate decline.
How to Combine Dynamic Friction and Fraud Scores
Fraud scores give businesses a highly configurable way to implement dynamic friction. Companies can tailor the scoring rules to their industry sector and continually tweak and monitor the results.
Fraud prevention systems often use machine learning in order to continually refine their scoring models. Many systems use a “whitebox” system, where businesses can proactively review how scores are constructed, effectively allowing them to unravel automated decisions.
This can assist both in tweaking the system for better balance in the future, and provide justification to customers in cases where false positives still occur even in spite of people’s best intentions.
The ideal endgame scenario is that genuine and loyal customers are completely unaware that their transactions are subject to any friction at all. Customers will only be aware of the other measures, checks and authentication procedures when a fraud score threshold is breached. If the system is sufficiently well-tuned, those “customers” may well be fraudsters – who the business isn’t concerned with “insulting” anyway.
The Dynamism of Dynamic Friction
With fraud trends evolving all the time, businesses have no option but to continually bolster and refine their fraud prevention mechanisms. But however they operate, one thing they don’t want to do is to offend genuine customers, which can potentially have even more financial impact than fraud itself.
Online businesses already have to operate in a world where, according to Shopify, 69.57% of online shopping carts are abandoned. Dynamic friction can reduce incidents of alienating the remaining 30.43% of people who really do want to hand over their money.
