Secure your business from login to chargeback
Stop fraud, break down data silos, and lower friction with Sift.
- Achieve up to 285% ROI
- Increase user acceptance rates up to 99%
- Drop time spent on manual review up to 80%
By Sarah Beldo /
Updated
We’ve got data …. it’s multiplying …
Seriously. In the past year, we collected more than 1.2 billion unique events each month, including identity, behavioral, and and network signals – all in the service of preventing fraud and other bad behavior online. With an incredible range of businesses sending information to Sift Science, we can learn insights on countless industry verticals and geographies. The more data we leverage, the more accurate and valuable our service is for our customers.
Among all that data, we also discovered some intriguing findings that we didn’t expect. For example, did you know that…
This finding might evoke a chuckle or two, considering that social network’s current reputation. But perhaps there’s a good reason that MySpace users have a higher fraud rate – it’s really, really easy to create an account. So, if you’re trying to piece together a fake identity quickly, you may turn to MySpace.
We also found that people with content-based social media accounts are less likely to be fraudsters, probably because it takes more effort to create these types of profiles. For example, people with Flickr or Vimeo accounts are 30% less risky than those without.
Sound counter-intuitive? Well, when you consider that order volume went way up on Cyber Monday, you’ve got part of the puzzle. Shoppers flocked online, and our data showed that the majority were legitimate customers making legitimate purchases.
And then consider fraud attempts didn’t spike on November 30 – they stayed steady. This makes sense if you think about online fraud as a highly organized “business” that is more like a job than an occasional hobby. People don’t suddenly decide to commit fraud, or to commit more fraud, simply because it’s a busy shopping day. So the proportion of fraudulent orders (steady) out of a total of all orders (an increased amount) was smaller that day.
We started our research into the Fraudiest States in America with some assumptions about where fraud was taking place. While some assumptions proved true – like discovering that the majority of fraudulent orders were being shipped to the Atlantic coast, likely to be reshipped abroad – others did not.
For example, not all major cities had lots of fraud. San Francisco, Boston, Dallas, and Chicago all had a lower fraud rate than average. And there were several examples of smaller or midsize cities (like Saginaw, Michigan) that had a higher fraud rate than much larger cities in the same state (like Detroit, Michigan).
There are some obvious fraud clues – like dozens of accounts created from the same IP address – that don’t require sophisticated algorithms to detect. But Sift Science specializes in spotting the subtle and nuanced patterns that could indicate fraud, which emerge when you combine multiple signals.
Digits are just one of the many fraud signals contained in an email address, and the fraud fact above was true as of June 17 (though exact stats are continually changing based on data and evolving fraud trends…)
Here’s another finding that raised some eyebrows when we first published it a few months back. However, it makes sense when you think about how fraudsters operate – they aim to look and act like a “normal” user and fly under the radar. Perhaps some criminals believe they’ll arouse less suspicion masquerading as an elderly person.
What will we learn about fraud in 2016? We can’t wait to find out…
Sarah Beldo was the Director of Content Marketing at Sift.
Stop fraud, break down data silos, and lower friction with Sift.