Dataiku Case Studies Online Fraud Detection

Edit This Case Study Record

	Online Fraud Detection Dataiku

Online Fraud Detection

Dataiku

Technology Category	Analytics & Modeling - Predictive Analytics
Applicable Functions	Sales & Marketing
Use Cases	Fraud Detection Predictive Quality Analytics
Services	Data Science Services
Challenge	SendinBlue, a relationship marketing SaaS solution, faced a significant challenge in validating new customers and ensuring the quality of their databases. The company had to ensure that all contacts on the list were opted in, which required manual validation. This process was not only time-consuming and required a large workforce, but it also severely delayed account validation for customers, damaging SendinBlue’s reputation. As the customer base grew, manual validation became increasingly unfeasible. The company needed a solution that could automate the validation process and scale with the growing demand. Read More
About Customer	SendinBlue is a relationship marketing SaaS solution that was launched in 2012. The company's mission is to become the simplest, most reliable, and cost-effective marketing platform. SendinBlue now powers marketing campaigns for more than 50,000 companies around the world. With offices in North America, Europe, and Asia, the SendinBlue team supports the product in six languages. Their platform integrates with the top e-commerce and CMS tools, and their system delivers over 30 million emails and text messages per day. Read More
Solution	SendinBlue turned to Dataiku Data Science Studio (DSS) to develop an automated fraud detection system. Using historical data from over 1 billion emails and associated events, thousands of blocked accounts, and hundreds of fraud criteria, SendinBlue built a scalable solution. The new system analyzes new customers and automatically classifies them as 'good,' 'bad,' or 'uncertain.' An algorithm then determines the customer’s credibility by taking into account sending volume, the scoring of the contacts, etc. Depending on the customer’s risk score, they may be blocked, validated, or sent to customer care for manual analysis. Dataiku was instrumental in deploying the data product into production, handling large amounts of different datasets, and designing, testing, and developing the solution in less than three months. Read More Log in to view content
Contents

Technology Category

Analytics & Modeling - Predictive Analytics

Applicable Functions

Sales & Marketing

Use Cases

Fraud Detection

Predictive Quality Analytics

Services

Data Science Services

Challenge

SendinBlue, a relationship marketing SaaS solution, faced a significant challenge in validating new customers and ensuring the quality of their databases. The company had to ensure that all contacts on the list were opted in, which required manual validation. This process was not only time-consuming and required a large workforce, but it also severely delayed account validation for customers, damaging SendinBlue’s reputation. As the customer base grew, manual validation became increasingly unfeasible. The company needed a solution that could automate the validation process and scale with the growing demand.

About Customer

SendinBlue is a relationship marketing SaaS solution that was launched in 2012. The company's mission is to become the simplest, most reliable, and cost-effective marketing platform. SendinBlue now powers marketing campaigns for more than 50,000 companies around the world. With offices in North America, Europe, and Asia, the SendinBlue team supports the product in six languages. Their platform integrates with the top e-commerce and CMS tools, and their system delivers over 30 million emails and text messages per day.

Solution

SendinBlue turned to Dataiku Data Science Studio (DSS) to develop an automated fraud detection system. Using historical data from over 1 billion emails and associated events, thousands of blocked accounts, and hundreds of fraud criteria, SendinBlue built a scalable solution. The new system analyzes new customers and automatically classifies them as 'good,' 'bad,' or 'uncertain.' An algorithm then determines the customer’s credibility by taking into account sending volume, the scoring of the contacts, etc. Depending on the customer’s risk score, they may be blocked, validated, or sent to customer care for manual analysis. Dataiku was instrumental in deploying the data product into production, handling large amounts of different datasets, and designing, testing, and developing the solution in less than three months.

Impact #1	The analytics initiative saved the equivalent of a full-time job and rationalized the global validation process, allowing SendinBlue to scale its team and customer growth exponentially.
Impact #2	SendinBlue was able to provide a better customer experience by considerably shortening validation delays.
Impact #3	The machine learning model now handles most of the new accounts, initially handling just 24 percent of the new accounts as part of a staged rollout.

Impact #1

The analytics initiative saved the equivalent of a full-time job and rationalized the global validation process, allowing SendinBlue to scale its team and customer growth exponentially.

Impact #2

SendinBlue was able to provide a better customer experience by considerably shortening validation delays.

Impact #3

The machine learning model now handles most of the new accounts, initially handling just 24 percent of the new accounts as part of a staged rollout.

Benefit #1	The first iteration of their model leverages Random Forest and has an 83 percent precision rate on email address classification.
Benefit #2	The machine learning model now handles most of the new accounts, up from initially handling just 24 percent of the new accounts.

Benefit #1

The first iteration of their model leverages Random Forest and has an 83 percent precision rate on email address classification.

Benefit #2

The machine learning model now handles most of the new accounts, up from initially handling just 24 percent of the new accounts.

Download PDF Version

Overview

Online Fraud Detection

Operational Impact

Quantitative Benefit