There is no one silver bullet to battle fraud. In our 11th
Annual Fraud Survey, while merchants
rate some detection tools as more helpful than others, it is possible that any given tool can be very useful for one type
of merchant but less effective for other
merchants. We believe the whole is
greater than the sum of the parts when it comes to deploying fraud tools,
especially when deployed in all four dimensions of fraud detection: Validation
Services, Merchant Proprietary Data, Multi-Merchant Data and Device Tracing Services.
The real strength behind the variety of tests is realized
when your system can:
1. Combine detectors from across the four dimensions to
pinpoint your fraud attempts (which reduces false positives), and
2. Apply different "sets" of rules to transactions with some
unifying characteristic (such as express shipments, or orders from long-term
customers). Compound, or multi-condition, rules allows a well-rounded approach
to focus on your specific fraud. Providing multiple rule sets (or “profiles”)
gives you a way to tailor the order screening “personality” to match the
expected fraud from specific groupings of your orders.
One word of caution: when you add new detectors or make
decision rule changes to your automated screening system, you should run these changes
in parallel instead of pushing them into production so that potential impacts
to the system (increased order rejection, or out sort for manual review etc.)
can be evaluated before full deployment.
Some merchants will run simulation with test files and test systems but
these simulations are less effective (or at the very least, less “lifelike”)
than using actual transaction data in real time, which also provides for an
accurate, apples-to-apples comparison.
In my next post, I’ll discuss how to find out if your
detection tools are integrated for optimization. If you want to read the 11th
Annual Fraud Report, download after registration.
Comments