Catch more fraud with machine learning.

Machine learning is a natural complement to human intelligence. SureScore is trained on millions of fraud records submitted by fraud experts. Our adaptive, real-time algorithm scores transactions and prevents fraud you might otherwise miss.

Sure Score 01 animated

Avoid unnecessary reviews.

Reduce your review queue and avoid costly step-ups by identifying low-risk transactions. Predictive machine learning shows which connections are fraudulent and which can be trusted.

Sure Score 02 animated

Identify good customers.

Predictive machine learning-based insights allow you to treat new customers like a well-known VIP. Make it easier for them to open an account, log in, make a purchase or take advantage of a promotion.

Sure Score 03 animated

Here’s how SureScore works.

SureScore, our powerful machine learning platform, monitors transactions and activity patterns that predict fraudulent behavior. Our models analyze hundreds of device attributes and are trained on millions of confirmed fraud reports to deliver a single risk score.

Assess Transaction Risk

SureScore analyzes confirmed fraud reports submitted by thousands of fraud experts to detect and score transaction risk.

Recognize Good Customers

Instantly identify good consumers to streamline their online experience even if they are new to you.

Real-Time Insights

We assess risk and deliver a score in real time, unlike many machine learning solutions that only work in batch mode.

Complement Rules-Based Systems

SureScore works together with rules-based systems to increase fraud catch and identify trustworthy consumers.

Streamline Fraud Operations

SureScore recognizes more trustworthy transactions than a rules-based system alone. Reduces manual reviews and the need for step-up authentication or call center verification.

Proof of performance.

Our customers are very important to us, and knowing who’s good makes all the difference in protecting and enhancing their experience with Paf.

Jessica Hård af Segerstad Data Analyst, Paf

Here’s what you get with SureScore Trust Indicator.

Our customer success specialists are ready to help you quickly integrate SureScore into our fraud prevention solution, and customize it for your needs.

  • Fraud prevention in real time: In about 100ms we are able to recognize a device, determine if it’s evading detection and associated with past fraud, and run all of the risk checks you’ve specified.
  • 99.9% uptime: Our distributed SaaS infrastructure supports some of the largest transaction volumes in the fraud prevention solutions business with no interruptions during updates or maintenance.
  • Quickly adapt to changing fraud patterns: Our powerful and flexible business rules editor enables fraud analysts to react immediately to new threats.
  • World-class fraud experts: Add our trusted fraud advisors to your team. Our customer success team partners with you to solve your unique business challenges and adapt to an ever-changing fraud environment.
  • Equipped for success: We provide continued training on new features and functionality through our help portal, eLearning webinars and our active user community.

Questions? We have answers.

Does SureScore use supervised or unsupervised machine learning?

SureScore uses supervised machine learning. SureScore was developed through a variant of the well-established Cross Industry Standard Process for Data Mining (CRISP-DM). CRISP-DM defines iterated steps of a common-sense approach to model development and deployment. SureScore is trained on confirmed fraud evidence submitted by your business and your peers in the FraudForce network.

What is SureScore actually predicting?

Generally, we talk about SureScore predicting risk and trust. Risk and trust are opposite sides of the same coin. If I have high trust then there is low risk, and if I have high risk then there is low trust. SureScore uses fraud evidence related to criminal behavior to represent "risk" in the model. Simply put, SureScore predicts the likelihood that a given transaction is fraudulent or trustworthy.

Is SureScore a batch or real-time process?

SureScore is 100% real-time. Not only are the transactions scored in real-time, but the attributes are also calculated in real time. Most machine learning is performed in batches, which is a much slower process and not suitable for latency-sensitive transactions. To provide the most robust and relevant information you need to frequently update the model statistics and attributes. SureScore works "hand in hand" with our expert rules-based fraud prevention solution, FraudForce, and returns your result in milliseconds.

What do the SureScore values mean?

For the transaction being processed, SureScore returns a value between -10,000 and +10,000. Generally, lower SureScore values are more risky, and higher SureScore values are more trustworthy. Analysis of your transaction and fraud placement behavior is the best way to understand how to apply SureScore to best suit your fraud prevention needs. Our Customer Success team is expert in assisting our subscribers in establishing meaningful SureScore targets.

What is the difference between a FraudForce check and a SureScore check?

FraudForce is an expert fraud prevention solution that provides a comprehensive rules management platform that allows customers to customize rules to their unique business requirements. SureScore is a machine learning analytic system that predicts the likelihood that the transaction under assessment will turn out to be fraudulent. It also predicts the likelihood that a transaction will turn out to be trusted, so you can identify more good customer transactions and reduce step-up authentication requests.

Our experts weigh in on SureScore, authentication, fraud and more.

See SureScore in Action

In just minutes, we’ll show you how to improve your customer authentication experience, stop fraud and save money.