Sift Case Studies Safely sending millions of dollars overseas every month

Edit This Case Study Record

	Safely sending millions of dollars overseas every month Sift

Safely sending millions of dollars overseas every month

Sift

Technology Category	Analytics & Modeling - Machine Learning Application Infrastructure & Middleware - API Integration & Management
Applicable Industries	Finance & Insurance
Applicable Functions	Business Operation
Use Cases	Fraud Detection
Services	Data Science Services
Challenge	Remitly, a company that enables customers to transfer money internationally, was facing a significant challenge with fraud. As the company expanded its services in the Philippines, fraud activity increased. The company's risk team, despite having several years of fraud investigator experience, was overwhelmed with the volume of data they had to review. They had set up 80-100 rules to capture suspicious activity, but scaling was proving to be a challenge. The system they were using was also incapable of learning from the company's data. Read More
About Customer	Remitly is a company that provides a platform for customers to transfer money internationally. The company's mission is to build modern solutions that empower people to send money and interact more personally. Customers can send money to their loved ones back home, all online and using their mobile phone. Money can be sent or picked up at over 10,000 locations or deposited directly into a bank – all in under two minutes. The company offers a bold 100% money back guarantee, which requires them to ensure that each transfer is reviewed systematically with rigorous investigation. Read More
Solution	Remitly integrated Sift's score API into its review workflow. Using the Insights Dashboard, Remitly settled on a risk score threshold that was right for them. They also used Sift’s customizable alerts to add review tasks in the work queue. The Sift Score helped the team tip the scale in the decision making process. Over time, the scoring mechanism started to detect more nuanced fraud patterns. Remitly was able to rely less on its complicated rules engine and started to see improvement. Nate Spanier, Director of Operations at Remitly, emphasized that in order to really get the maximum benefits of integrating with Sift, customers will need to commit to managing their data. Read More Log in to view content
Contents

Technology Category

Analytics & Modeling - Machine Learning

Application Infrastructure & Middleware - API Integration & Management

Applicable Industries

Finance & Insurance

Applicable Functions

Business Operation

Use Cases

Fraud Detection

Services

Data Science Services

Challenge

Remitly, a company that enables customers to transfer money internationally, was facing a significant challenge with fraud. As the company expanded its services in the Philippines, fraud activity increased. The company's risk team, despite having several years of fraud investigator experience, was overwhelmed with the volume of data they had to review. They had set up 80-100 rules to capture suspicious activity, but scaling was proving to be a challenge. The system they were using was also incapable of learning from the company's data.

About Customer

Remitly is a company that provides a platform for customers to transfer money internationally. The company's mission is to build modern solutions that empower people to send money and interact more personally. Customers can send money to their loved ones back home, all online and using their mobile phone. Money can be sent or picked up at over 10,000 locations or deposited directly into a bank – all in under two minutes. The company offers a bold 100% money back guarantee, which requires them to ensure that each transfer is reviewed systematically with rigorous investigation.

Solution

Remitly integrated Sift's score API into its review workflow. Using the Insights Dashboard, Remitly settled on a risk score threshold that was right for them. They also used Sift’s customizable alerts to add review tasks in the work queue. The Sift Score helped the team tip the scale in the decision making process. Over time, the scoring mechanism started to detect more nuanced fraud patterns. Remitly was able to rely less on its complicated rules engine and started to see improvement. Nate Spanier, Director of Operations at Remitly, emphasized that in order to really get the maximum benefits of integrating with Sift, customers will need to commit to managing their data.

Impact #1	Remitly was able to quickly integrate the API in an afternoon and start teaching Sift’s machine learning models with Remitly’s data.
Impact #2	The scoring mechanism started to detect more nuanced fraud patterns over time.
Impact #3	Remitly was able to rely less on its complicated rules engine and started to see improvement.

Impact #1

Remitly was able to quickly integrate the API in an afternoon and start teaching Sift’s machine learning models with Remitly’s data.

Impact #2

The scoring mechanism started to detect more nuanced fraud patterns over time.

Impact #3

Remitly was able to rely less on its complicated rules engine and started to see improvement.

Benefit #1	50% Fewer manual reviews
Benefit #2	20% Reduction in new user reviews
Benefit #3	70% of fraud is caught in the top 2% of transactions

Benefit #1

50% Fewer manual reviews

Benefit #2

20% Reduction in new user reviews

Benefit #3

70% of fraud is caught in the top 2% of transactions

Download PDF Version

Overview

Safely sending millions of dollars overseas every month

Operational Impact

Quantitative Benefit