Fraudsters come in many sizes (individuals with a stolen purse, organized rings of credit card thieves, evil bots) and with many habits. In last week’s original “10 Warning Signs of eCommerce Fraud”, our list covered a wide range of indicators. In this post, we’ll cover the first three signals: First time shoppers, bulk orders, and order variance. More often than not, signals indicating a first-time shopper who deserves a second glance.

man@work 343/365 by Dennis Skley, on Flickr
man@work 343/365 by Dennis Skley, on Flickr

Apparent first-time shoppers are risky, because there’s not enough data to confidently determine that that customer legitimately interacted with your site in the past. Criminals are always on the prowl for new victims. Generally, fraudsters are unwilling to wait weeks or months to begin stealing with an account that started off as legitimate. They operate with the knowledge that once their accounts are used to create bad orders, those accounts can no longer be used on the same site. Additionally, merchants should be wary of bulk orders since they could be indicative of purchases with a stolen credit card. In order to maximize their potential profits before credit cards are reported missing or hacked accounts are noticed, criminals may order lots of items from a website rapidly.

We all probably wish that we could order 5 different Rolex watches or 6 different styles of Louis Vuitton bags, just to try out which look best (or to keep). However, when these kinds of high-end goods are purchased in both a quantity and variety outside of the average shopper’s range, fraud may be present.

Criminals may attempt to capitalize on bulk ordering while also profiting from a single account — rather than creating a new user on a new website — by ordering various iterations of the same item. Especially for high-value goods, minimizing the opportunity for notice is key for fraudsters. These orders in-bulk can be a red flag for fraud detection teams.

New accounts — especially those quickly created, not linked to historic transactions or credit cards, and used to purchase large orders — are therefore suspicious.

Of course, not every bulk order or new shopper is bad. Whether a shopper is simply trying to complete all of his holiday shopping before the rush or plans on “swagging out” his first softball team, merchants should be certain to verify order details before canceling a transaction outright.

While setting up a simple rule might sound like the solution, blocking all first-time shoppers’ orders will also cancel any potential good users’ transactions. That’s where machine learning comes into the equation.

With a machine learning-based system like Sift Science, merchants can train their fraud detection algorithms to recognize first-time shoppers while weighing the shoppers’ relative newness against thousands of other indicators. These indicators — signals, in Sift terminology — combine to offer a holistic view of shoppers, providing a bigger-picture understanding of accounts and their users.

Long story short: save time reviewing first-time shoppers with Sift Science; they may be fraudulent, but you won’t know for sure unless you have the whole picture.

Thanks for tuning into Part 2 of our “10 Warning Signs of eCommerce Fraud” series. For more information about machine learning and fraud, visit our free Guides and Resources library or tweet us questions @siftscience.