At Sift Science, we know how intense the holiday period can be for retailers. Many of our e-commerce customers rely on our machine learning-based fraud detection solution to help them stay sane amidst a significant spike in order volume. The more aspects of fraud management that can be automated, the less burden on your internal review team. High fives all around!
This year’s Cyber Monday was huge for merchants, with more than $3.1 billion spent online in a single day! That made us wonder: were fraudsters also taking out their (stolen) credit cards that day?
Surprising, but true: Fraud didn’t go up
We took a sample of 1 million items from orders placed on Cyber Monday, then applied a commonly used risk threshold (basically, the point at which a typical e-commerce merchant would block an order) to determine which ones were fraudulent.
After we dug into our data, we were pleased to see that our customers enjoyed a very jolly Cyber Monday. The number of orders created that day was 30% higher, compared with the rest of November.
We were equally pleased – though also somewhat surprised – to see that overall fraud rates remained steady on November 30. In fact, if you look at the relatively consistent volume of fraudulent orders as a proportion of total orders (of which there was a significant jump), you wind up with a lower fraud rate on Cyber Monday than on a typical day.
This supports a hypothesis you may have heard before: online fraud isn’t opportunistic, it’s organized and calculated. Someone doesn’t just decide to make a fraudulent purchase on Cyber Monday; it’s just another day in what’s (basically) a job.
Knowing that there wasn’t a deluge of fraud attempts on this busy shopping day could give merchants a clue for how to adjust their anti-fraud strategy. For example, since legitimate orders are a larger proportion of the whole, they may want to focus less on identifying fraudulent orders, and more on making sure they’re letting as many good customers through as possible. This could mean re-evaulating the threshold for what they review, or reducing friction at checkout for users who they know aren’t risky.
Fast fraud: digital goods are easy targets
When looking at fraud attempts on Cyber Monday, we also looked at individual categories of items and different brands, to see if there were patterns. We found that two categories in particular had much higher fraud rates than others: gift cards and gaming.
Both gifts cards and gaming share some common traits that make them ripe fraud targets. In essence, they’re digital cash, which doesn’t require a shipping address to place an order. Not only does this make digital goods more attractive, since a fraudster doesn’t have to arrange reshipping and reselling, but it also makes it harder for businesses to tell whether that particular order is legitimate or fraudulent.
And, since customers are expecting their purchases to arrive in mere seconds, merchants don’t have time to review each order. (That’s where an accurate, real-time fraud-detection system really comes in handy…)
What brands do fraudsters target?
We thought it could be interesting to see if fraudsters go after one particular brand over another, particularly when it comes to legendary rivals. Here’s some of what we found:
Playstation or X-box? Devices and gift cards related to these platforms are about equally likely to be targets of fraud.
Coke or Pepsi? Coke wins with both legitimate shoppers and fraudsters. A higher volume of Coca-Cola products than Pepsi products were sold on Cyber Monday – and Coke also had a higher fraud rate.
Nike or Adidas? Fraudsters appear equally likely to steal both brands of shoe.
Apple or Samsung? Samsung products popped up more often when it came to attempted fraud, but accessories related to the Apple Watch – like docks and stands – were also popular.