The Most Accurate Fraud Detection in the Industry
97%
transaction approval rate
95%
fraud detection
<5%
intervention

Built to Detect Online Fraud

The AI-Powered Outseer Risk Engine is built to detect online fraud with industry leading precision. On the strength of over two decades of experience in artificial intelligence and machine learning, Outseer has developed highly effective predictive AI models that continuously learn frompast behavior to identify fraudulent behavior in real-time with proven accuracy.
Predictable challenge rate based on your risk appetite
Our normalized risk score gives you the predictability you need to choose the risk thresholds that meet your business objectives. You can balance your fraud risk appetite with customer intervention levels to control the amount of friction your customers experience; the risk policies you set will determine whether a user is challenged with a step-up authentication. By applying a risk-based approach of your choice, only a small number of activities or transactions require additional authentication so you only challenge customers when necessary and at a predictable rate.
Unique customer scoring and explainable results
The Outseer Risk Engine tunes the weightings of the risk model predictors for each unique customer based on the feedback from case resolution and authentication results; this ensures high risk score accuracy to quickly detect emerging fraud patterns. Unlike black-box risk engines, up to ten contributors that impacted the risk score for a transaction are shown in Case Manager for explainable results, ensuring an informed case review and decision-making process.
Orchestrates first- and third-party fraud signals
The Outseer Risk Engine orchestrates and ingests data signals and intelligence from your systems, as well as third-party fraud signals. This data signal orchestration and ingestion ultimately drives the best possible risk decision on each event.
consortium data signals

Orchestration and Analysis of Superior Data

Our Risk Engine works in real-time to orchestrate and analyze a wide range of data signals associated with an activity or transaction as well as first- and third-party signals, to determine the probability of a given interaction being genuine or fraudulent. The Outseer Risk Engine further enriches the data set with essential inputs from our superior consortium data signals from the Outseer Global Data Network.
 
The use of our unique global consortium data generated across Outseer Fraud Manager and Outseer 3-D Secure gives customers insights into fraud that would be impossible to otherwise obtain. With these broad global data inputs, you can identify known and emerging threats across digital channels with the highest accuracy, all while reducing friction for legitimate customers and transactions.
risk engine meter

Highly Accurate Scoring with Advanced AI and ML Algorithms

The Risk Engine uses an advanced statistical approach to calculate the conditional probability of an activity or transaction being fraudulent, based on the properties of the transaction, case marking from the Case Management tool, and authentication feedback provided by the customer. The Risk Engine uses the right data, at the right time for the right use case, resulting in high performance and a superior ability to explain what is truly driving your risk score.
 
The score is normalized to create a usable predictive risk score that accurately assesses the risk associated with each transaction and digital interaction during the user journey. The end result is a comprehensive risk profile uniquely tailored to each customer.

Try Our Platform of Solutions

Request a Demo

The Risk Scoring Process

The process starts when a user makes a transaction, logs in or carries out a post login activity such as updating their profile.
report icon
1) Facts Gathering
The Risk Engine collects raw data facts from diverse inputs including user behavior from past and present data, the device used, data from the Global Data Network, and anomaly detection used while analyzing the risk of the transaction. These signals are enriched and organized using a uniform data format and are then optimized to analyze the risk presented by a transaction.
checklist icon
4) Preliminary Risk Score
The Outseer Risk Engine generates a preliminary risk score for each transaction based on the probability of fraud given the predictors present in the transaction. All data variables are taken into account in this preliminary score, with each one weighted based on its relevance. This process also produces a list of indicators to explain the major factors influencing the score.
tech architecture icon
2) Profile Building
Individual customer profiles are built using this data, along with sets of accumulated historical data and statistics about the user such as IP age to that account, cumulative payment amounts, payment velocity, and other relevant data. By using these profiles, the Outseer Risk Engine can distinguish between normal transaction behavior and abnormal behavior that may suggest fraud.
solution brief icon
5) Risk Score Normalization
The risk score generated is then normalized to a logarithmic scale from 0 to 1,000. The final scores represent a fraud likelihood and provide predictability and control over the percentage of friction caused on a “normal” daily basis. This normalization provides more control over the percentage of transactions requiring intervention.
data network icon
3) Predictor Generation
Predictors, or data variables in the risk model known to correlate with fraud, are generated from the facts about the current transaction or calculated from historical data from profiles such as IP country age. They can be end-user dependent or general population dependent. The better the predictors, the more accurate the prediction of fraud.
legal icon
6) Adjust Weights & Self Learning
The final step of the Risk Engine scoring process is adjusting weights with self-learning. The Risk Engine learns automatically and re-calculates the weightings of the risk predictors. The weights are dynamically tuned per customer, daily. The risk scoring accuracy improves along with the system use in terms of historical reference and feedback to the Risk Engine.
report icon
1) Facts Gathering
The Risk Engine collects raw data facts from diverse inputs including user behavior from past and present data, the device used, data from the Global Data Network, and anomaly detection used while analyzing the risk of the transaction. These signals are enriched and organized using a uniform data format and are then optimized to analyze the risk presented by a transaction.
tech architecture icon
2) Profile Building
Individual customer profiles are built using this data, along with sets of accumulated historical data and statistics about the user such as IP age to that account, cumulative payment amounts, payment velocity, and other relevant data. By using these profiles, the Outseer Risk Engine can distinguish between normal transaction behavior and abnormal behavior that may suggest fraud.
data network icon
3) Predictor Generation
Predictors, or data variables in the risk model known to correlate with fraud, are generated from the facts about the current transaction or calculated from historical data from profiles such as IP country age. They can be end-user dependent or general population dependent. The better the predictors, the more accurate the prediction of fraud.
checklist icon
4) Preliminary Risk Score
The Outseer Risk Engine generates a preliminary risk score for each transaction based on the probability of fraud given the predictors present in the transaction. All data variables are taken into account in this preliminary score, with each one weighted based on its relevance. This process also produces a list of indicators to explain the major factors influencing the score.
solution brief icon
5) Risk Score Normalization
The risk score generated is then normalized to a logarithmic scale from 0 to 1,000. The final scores represent a fraud likelihood and provide predictability and control over the percentage of friction caused on a “normal” daily basis. This normalization provides more control over the percentage of transactions requiring intervention.
legal icon
6) Adjust Weights & Self Learning
The final step of the Risk Engine scoring process is adjusting weights with self-learning. The Risk Engine learns automatically and re-calculates the weightings of the risk predictors. The weights are dynamically tuned per customer, daily. The risk scoring accuracy improves along with the system use in terms of historical reference and feedback to the Risk Engine.

The Outseer Platform

The Outseer Risk Engine, together with our superior data from the Outseer Global Data Network, Policy Manager and Case Manager, comprise the Outseer Platform. The Platform enables our customers to provide comprehensive fraud protection across the digital journey, in contrast to point solutions that are focused on narrow use cases and are often missing policy and case management componentry.
outseer platform flow diagram
The Global Data Network powers the Risk Engine
The Outseer Global Data Network is a collaborative data consortium that aggregates signals from millions of transactions across thousands of the largest financial institutions worldwide. By using these shared network signals, financial institutions can proactively identify fraud or attempted fraud.
The Risk Engine creates a risk score
The Risk Engine takes the data signals from the digital transaction, along with first- and third-party data, and data from the Global Data network, to assign a preliminary risk score. This score is normalized, providing your team with predictable intervention rates.
The score informs a decision made through the Policy Manager
The Policy Manager decisions the normalized score to determine if this transaction should be allowed, warrants a step-up authentication, or should be denied. All of these steps occur in milliseconds, allowing your customer to have a seamless experience, while still being protected from fraud.
Transactions can be reviewed in the Case Manager
Decisions made by Policy Manager to identify potential fraud, can be reviewed in Outseer’s Case Manager, where the risk model publishes up to ten AI explainability indicators for the score produced for each transaction, showing the signals that contributed most to the final score, along with a summary of some of the most relevant data points.
The Global Data Network powers the Risk Engine
The Outseer Global Data Network is a collaborative data consortium that aggregates signals from millions of transactions across thousands of the largest financial institutions worldwide. By using these shared network signals, financial institutions can proactively identify fraud or attempted fraud.
The Risk Engine creates a risk score
The Risk Engine takes the data signals from the digital transaction, along with first- and third-party data, and data from the Global Data network, to assign a preliminary risk score. This score is normalized, providing your team with predictable intervention rates.
The score informs a decision made through the Policy Manager
The Policy Manager decisions the normalized score to determine if this transaction should be allowed, warrants a step-up authentication, or should be denied. All of these steps occur in milliseconds, allowing your customer to have a seamless experience, while still being protected from fraud.
Transactions can be reviewed in the Case Manager
Decisions made by Policy Manager to identify potential fraud, can be reviewed in Outseer’s Case Manager, where the risk model publishes up to ten AI explainability indicators for the score produced for each transaction, showing the signals that contributed most to the final score, along with a summary of some of the most relevant data points.