Page 26 - GS161202
P. 26
Views
"It might be better to call it fraud minimization, because lect fraud prevention tools, but artificial intelligence
you'll never eliminate 100 percent of ecommerce fraud," will never replace human oversight.
he said. "And using machine-learned, rules-based systems
to optimize a website may inadvertently penalize good When numbers got personal
customers who don't fit the profile. For example, if a
traveler in California wants to send a present to his wife Dakshina warned against fraud prevention tools that cost
in New York, the transaction may be rejected due to more than losses due to fraud. He advises small business
mismatched billing and shipping addresses." owners to use human oversight until their transaction
volumes justify the investment in automated prevention
The best fraud mitigation methods are sophisticated and tools. He also advises them to review merchant services
nuanced, using an arsenal of tools, each with distinct applications to understand fees and pricing and protect
pros and cons, Dakshina added. He cited the following their personal information.
examples:
"We evaluate and prequalify some merchants on behalf of
• Human oversight: This traditional method is supe- ISO partners," Dakshina said. "We upload their documents
rior to machine learning at detecting fraud but lacks in a secure server and directly forward to their processor.
the ability to scale. When you have high volumes of It's a value-added service that makes the underwriters'
orders, the tool becomes obsolete. jobs easier."
• Rules-based fraud prevention tools: These tools This service and others like it have transformed merchant
can be useful in blocking certain types of fraud but services from transaction-based interactions into richly
require time and effort to be properly implemented. rewarding, creative relationships among merchants,
Fraudsters will do everything possible to see how customers and service providers.
they can break through.
Dale S. Laszig, Staff Writer at The Green Sheet and Managing Director
• Machine learning: This self-learning system can at DSL Direct LLC, is a payments industry journalist and content pro-
become highly proficient at identifying certain pat- vider. She can be reached at dale@dsldirectll.com and on Twitter at @
terns in fraud and evaluating the efficiency of se- DSLdirect.
