top of page

AI Case Study

Chime decreases basis point loss by 40% using a machine learning fraud detection platform

Chime online banking implemented Simility's machine learning fraud detection platform to identify fraudulent users and payments more quickly based on historic and live data, reducing loss basis points by 40%.

Industry

Financial Services

Banking

Project Overview

"The Simility solution provided a multi-pronged fraud prevention solution for Chime - ingesting Chime’s custom and extensible data model with historical and live-streaming data, and leveraging its powerful machine-learning models combined with a flexible rules engine. This enabled Chime to identify and block the fraudulent activity that had earlier gone undetected. But with Simility’s advanced device fingerprint technology, Chime could detect multiple accounts created from the same device and suspend them—before a single stolen payment was processed.

Additionally, all related activities by the fraudulent device were made available to the fraud analysts to catch unusual transaction patterns (such as distinct accounts linked or debit transfers) per device. Simility’s complete fraud prevention solution enabled Chime to ingest several unconventional data points and sources (including fraudulent members’ spending patterns and their number of deposits) with relative ease. Simility’s UI-based ontology manager helped define entity relationships and enrich the models to identify members connected to multiple devices as an important predictor of fraudulent transactions."

Reported Results

Chime was able to reduce basis points loss by 40%, from 18 basis points to 11.

Technology

Function

Risk

Audit

Background

"Chime is one of the leading mobile-based financial services targeted at millennials. It offers a mobile bank account with no overdraft or monthly fees." Due to its business model, Chime faces an increased risk of customer fraud: "In the past, fraudsters had used Chime’s customer accounts to funnel money via stolen identities."

Benefits

Data

Historic and real-time data, such as account spending and deposit data, along with other "unconventional" data

bottom of page