TeleSign has long lead the way in using end-user-provided phone numbers to help online and mobile application businesses increase their account security. With our Score API, we look at data, intelligence, traffic patterns and reported fraud on phone numbers (from all over the world) in order to deliver a fraud risk assessment score that helps determine whether a user should or should not be able to register for an account. But fraud is always evolving. Bad actors are resourceful and over time find new ways to circumvent existing security measures by improving their behavior patterns. It’s the mission of our product and engineering teams to stay ahead of these threats by offering our customers the most innovative and effective ways to defend against them. Our latest effort comes in the form of an update to Score that represents a first-of-its-kind fusion of machine learning and phone number analytics that more accurately distinguishes good users from bad at account registration and/or during high-risk transactions.
What Is Machine Learning?
Machine Learning is a data analysis method that enables an algorithm to use historical indicators to uncover hidden insights in data and better predict future events, without being explicitly programmed. It is an increasingly utilized technique that evolved from the study of pattern recognition and the theory that computers can learn without being programmed to perform specific tasks (artificial intelligence). It’s an area of computer science that was first defined in 1959 and today is helping companies in all industries enhance their products and services.
How Does TeleSign Use Machine Learning to Prevent Fraud?
To make the Score API extremely accurate in detecting potential fraud, TeleSign’s data scientists leverage labeled data (end-user phone numbers identified as “good” and “bad”). This data allows for customized models to be developed by customer, industry, or use case. Through iterative machine learning, these models are then applied to new, unlabeled data so that a prediction can be made about the risk of each end-user attempting to create an online or mobile app account. The models adapt as they are exposed to and learn from new data, enabling a better, faster and more accurate analysis.
This updated Score offering combines machine learning with the rules-based system that was already functioning in the API. Customers now benefit from the addition of predictive data derived from machine learning models to the massive volumes of real-time and historical data on phone number usage consistently available to TeleSign through deep relationships with mobile network operators and a patented database of customer-contributed phone number reputation information (TeleBureau™). Used and analyzed together, the result is a data-driven fraud risk assessment score that helps businesses decide whether to allow, block or flag (for further review or verification) a new user during account registration.
How Can My Business Get Started?
The Score API can be easily integrated into existing account security workflows to help companies combat a variety of high-impact fraud types including spamming of good users, fake listings, credit card fraud, phishing attacks and promo and coupon abuse. Score also helps companies streamline registration to responsibly grow their authentic user base (with higher usage stats and increased average revenue per user) and protect their brand reputation.
To learn more about how Score works, check out our infographic, here.
To get started today with Score, sign up for free, here.