Do You Need A Fraud Model?
We reviewed the nuts and bolts of how fraud models work and how they can provide your solution with lots of information. As we discussed in our identity fraud solution cost benefit analysis, many risk analysts still see fraud models as a waste. Let’s discuss some different approaches commonly used in place of a fraud model. The two most common approaches we see are:
· Running an authentication-only fraud solution after pulling credit data
· Triggering an authentication fraud solution based only on internal data
Approach I – Running an authentication-only fraud solution after pulling credit data
Advantages – The credit report is a required element of any credit request, and credit bureaus can append useful elements to help with the fraud determination.
Disadvantages – The primary disadvantage is cost. Credit reports are the most expensive part of a credit request. Pulling the credit report only to discard it is wasteful spending. A fraud model can give all of the same output that commonly gets appended to a credit request without the costly credit data. Another disadvantage is the effect on the customer experience. Relying on your credit pull as the sole fraud indicator will create many false positives that will require identity authentication. The invasive nature of this process will push away profitable customers unwilling to deal with an additional 15-30 minute headache.
Approach II – Triggering an authentication fraud solution based only on internal data
Advantages – Internal flags like channel and a possible promotion code are great indicators of potential fraud. Our analytics team has reported impressive gains when this data is included in modeling efforts. Internal data also serves as the cheapest form of segmentation, as there is neither a model cost nor a credit pull.
Disadvantages – Internal data doesn’t present a complete picture. It doesn’t benefit from fraud “hot lists” or other information managed by third party providers. It also doesn’t benefit from velocity monitors that catch cross-channel fraud. It lacks the predictiveness of a fraud model. A fraud model has all of these characteristics and serves as a better segmentation tool than internal data alone.
Best practice fraud solutions need a fraud model. What is your approach, and how does it compare to our examples?
If you want to talk to an Equifax specialist about your fraud solution you can e-mail us here
If you are interested in a monthly e-mails about blog updates, sign up for our newsletter
Rich Huffman is VP of Product Management for Equifax Technology and Analytical Services. Rich is responsible for driving the market strategy for Equifax’s portfolio of ID Verification and Authentication products.
Rich brings over 17 years of experience in product management to Equifax. Prior to joining Equifax, Rich created and managed products for the financial services industry at Harbinger, S1, and ADP. Rich is an expert in utilizing ID verification and authentication technologies in addressing Red Flag and other compliance concerns for online banking and online payment related activities.
Rich graduated from Clemson University with a B.S. in business with a concentration in economics.