Feature Spotlight: Suspicious Patterns in User Behavior
By Jonathan Hsieh /
8 Dec 2015
We love giving our users an opportunity to discover more about what’s going on behind the scenes. Sometimes we’ll hear one of our customers say, “Can machine learning really do that?” or “Wow, I didn’t realize that’s how Sift Science worked!”
So we’ve introduced a new blog series called Feature Spotlight to give you a peek under the hood of Sift Science and dig into the various features that make up our machine learning platform. In each post, we’ll examine one of the features that Sift Science analyzes to detect fraud – ranging from “Does the user share a Device Fingerprint with a known fraudster?” to “transaction velocity over the past day”. Check it out and send us your questions!
If you noticed someone strolling by your house one evening, you probably wouldn’t think too much about it. But what if you saw him walking by again later that night? What if you saw him peering into your neighbor’s window? By this point, you’d probably consider his behavior as being highly suspicious. Even though each of his actions in isolation might not seem particularly noteworthy, piecing together the events raises red flags that something suspicious might be happening.
At Sift Science, we keep an eye out for suspicious online behavior to protect your business. One of the thousands of signals that our fraud platform uses to detect fraud is user behavior patterns – the patterns of how someone navigates on your site. When a user is interacting with your site, they’re essentially leaving behind a navigational footprint; our machine learning technology analyzes this behavior and looks for patterns that are likely to be fraudulent.
What does this mean? Take a look at the following user behaviors to see how this works:
- Login → Click on Product #8473 → Click on Product #157 → Click on Product #102 → → Complete Purchase
- This series of navigation seems typical of a user who’s shopping around a website, subsequently selecting a product to purchase. Sift Science would recognize that this person is likely to be a legitimate customer.
- Failed Login → Request Password → Direct Link to Product #821 → Change Shipping Address →Complete Purchase
- There are a couple red flags that pop up in this user’s behavior. They requested a password after a failed login, they directly navigated to a product page, and they changed the default shipping address.
- While these events might not be suspicious on their own, Sift Science’s machine learning analysis is able to deduce that this pattern is more likely to be fraudulent (and lets you know immediately).
To determine whether a user’s behavior on your site is fraudulent or not, Sift Science analyzes thousands of different patterns and data points about a user’s interaction with your site. Not only are we able to detect suspicious behavior patterns for your business, but we also learn what behavior patterns are especially significant for your unique business and continue to improve over time.
Want to learn more? Stay tuned for our next installment of Feature Spotlight!