What Does Social Say?
By Emily Chin /
14 Apr 2015
How many Twitter followers do you have? Do you still have a live MySpace profile? When’s the last time that you uploaded a photo to Flickr? The answers to all of these questions provide clues to your authenticity as a “good” online user. But what are some of the indicators that social data can suggest about fraudsters?
Shoppers’ social networks — literally, the digital connections that they have among other internet users — offer great information on the historical online behavior of their account holders. According to our findings, shoppers with…
- Myspace accounts are 1.3X more risky than those without
- Twitter handles with zero followers are 2X more risky
- Flickr or Vimeo accounts are 30% less risky
From these data points, we can suggest a few patterns: 1) People with content-based social media accounts are less likely to be fraudsters; and 2) Fraudsters are less likely to put forth the effort necessary to create false accounts on content-based sites.
For fraudsters, social networking sites that are based on uploading, storing, or sharing content, such as Flickr — a photo sharing site — or Vimeo — a video sharing site — require too much time and creativity to dupe. Why? Because these sites tend to have a more intensive account creation process. Flickr’s account creation requires that a new user input his birthday, telephone number, and gender, a time-consuming process if creating multiple fake accounts is the goal.
Vimeo’s registration system requires a plan selected from a tiered pricing model, engaging the user and slowing the registration process. Additionally, the purpose of both of these sites is on the input (read: upload) of media. Why would a fraudster create an account on a media sharing site if that would require an extra step — creating and uploading media? On the other hand, the barrier to entry for Twitter and Myspace are lower, and require little-to-no engagement upon account creation. To create accounts on either site, a user simply needs a name and email address, which we know can be easily obtained.
Sift Science’s machine learning system is constantly learning from our global network of fraud analysts. Data like trends in the social media habits of fraudsters versus legitimate users offer deep insight into online account and transactional activity that can only come from our extensive and adaptive fraud solution.