A company’s ability to identify and prevent fraud has always been critical. But now, with COVID-19 forcing more commercial activity online than ever before, the need for effective fraud prevention has become even more evident. As more and more transactions and interactions go digital, fraud continues to rise and evolve. In total, cybercrime is projected to cost the world $6 trillion annually by 2021.
With so much money at stake, merchants everywhere are looking for the best fraud prevention solutions. However, it can be hard for merchants to know what kind of approach to fraud prevention will be most beneficial for their business. There are many options on the market and a variety of factors should be considered in order to make the right decision. To help merchants navigate this choice, Mercator Advisory Group partnered with Forter, a leading fraud prevention company, to publish a white paper on topic.
The paper identifies the common problems and pain points that legacy fraud prevention approaches create for merchants and offers recommendations on what capabilities an effective solution must include.
Legacy approaches aren’t enough
Many merchants have responded to fraud by adopting approaches that address specific fraud vectors in isolation (i.e. Account Takeover (ATO), coupon abuse, transaction fraud, etc.). This kind of siloed approach means that different stages of the consumer journey are dealt with in isolation. For example, one solution might be applied to transaction fraud while another might be applied to new account creation. This leaves the merchant with gaps between their tools – resulting in higher operational costs as a result of manual teams needed to manage multiple vendors and can lead to more inaccurate results.
The paper notes that legacy approaches to stopping fraud are riddled with problems. Since the solution doesn’t look at the entire consumer journey but instead only at specific aspects of the customer journey, fraudsters can easily exploit gaps in protection. This legacy approach of leveraging multiple tools to create the merchant fraud stack results in a lack of comprehensive understanding of the context behind disparate data points and the story behind the digital identity that may be on the merchant’s site.
Traditional approaches are also hard to scale. When shopping volume increases, as it does during the holidays, systems can struggle to keep up with increased demand. And when new forms of fraud emerge, the legacy systems often struggle to identify them.
By only looking at their own data, merchants are not able to proactively stop fraud or anticipate growing fraud vectors that may eventually strike their business. This restricts the merchant not only from being able to scale during peak periods, but likewise curbs their ability to expand their products and services into new markets or geographies as a result of risk aversion.
Merchants need an integrated platform across the entire purchasing journey
Instead of a siloed approach, Mercator recommends that companies embrace a comprehensive solution that couples machine learning with massive data sets and ongoing human fraud expertise and analysis. At the heart of this approach is the goal of verifying the digital identity of the user. A merchant needs to know who the user is and whether or not that user is trustworthy.
The white paper identified five capabilities a modern fraud prevention system needs:
- An integrated platform that provides protection across all consumer touch points in the purchasing journey
- A global data network
- Machine learning for greater accuracy
- Advanced fraud analytics
- Fraud models tailored to individual enterprises
The white paper explains that companies need a solution that takes into consideration the full sweep of the customer journey – from login, to coupon redemption, and beyond – using as much data and information as possible. The data should not just be sourced from a single merchant, or even a single merchant vertical, but instead from a global data network of merchants spanning industries and geographies.
The optimal fraud prevention approach should gather data from a wide global data network and be constantly curated by teams of advanced fraud experts. Machine learning—the technology underpinning the best fraud prevention techniques— is only as good as the data that the system is fed. The models cannot be left to themselves and be expected to yield accurate results. . A hybrid approach of man and machine learning is required in order to accurately identify fraud and abuse.
When models are trained from data collected from millions of data points across geographies and different merchant categories, they become more likely to reliably detect a greater range of fraud.
Put simply, the ideal fraud prevention solution should have access to the best data sets, be connected in one place, and be overseen by knowledgeable experts interpreting the data. While machine learning is a crucial component, it must be coupled with human experts who can tweak the algorithms and interpret the data accordingly. If this approach is done properly, and is able to be tailored to the specific business requirements of the merchant, the result will be improved approval rates, a reduction in false declines, and slashed costs in operational overhead.
To learn more about the best way forward in fraud prevention solutions, read the white paper here.
