ACH Risk Score
ML model for ACH return likelihood
NOTE: Please contact your CSM to join the waitlist.
Unit21 processes millions of ACH transactions across our customer base daily. At such volumes, the ability to screen ACH payments for risk factors is an integral element of our fraud solutions.
Our ACH Risk Score is built on a Machine Learning model that utilizes thousands of features from transactional data - trained specifically on ACH transactions, to predict the likelihood of an unauthorized return (R10).
We predict R10s based on data across the Unit21 network, but critically including custom data that is specific to your institution and the types of products and services each customer offers, allowing the model to be tailored to the vectors of ACH fraud that are most relevant.
The model produces a score from 0-100 per ACH transaction, allowing you to set thresholds appropriate for your business when defining the parameters.
How Does It Work?
Once the ACH Risk score model is deployed for an organization, every relevant transaction will include a risk score between 0-100. Based on the value, customers are able to specify Risk Segments
When deployed, the Risk score can be used as a criterion to escalate ACH transactions that have a higher likelihood of being unauthorized returns as Alerts in the Unit21 platform.
Since each Risk score is tailored to an organization, the ACH Risk Score scenario model helps provide further automation, so that only the highest-value transactions are placed in review. Please navigate to "Creating an ACH Risk Score" for further details.
Updated 26 days ago