SendinBlue Sends the Right Message with Dataiku
SendinBlue, a relationship marketing company, sends millions of marketing- and sales-related email and text messages each day on behalf of its more than 50,000 corporate clients around the world. Ensuring that those messages reach the right people is crucial, and the company also has its reputation to consider.
Since launching as a company in 2012, SendinBlue has been steadily gaining momentum, says Jules Jeanroy, the company’s deliverability expert. “Our platform has welcomed more than 600,000 users across 140 countries to date, and with this growth has also come the need to more efficiently and effectively scale our operations. We needed to find a way to automate the time-intensive, manual task of validating new customers as approved or unapproved SendinBlue accounts that have permission to send messages from our platform.”
The Paris-based company which maintains U.S. operations in Seattle, also needed a solution that included machine learning to manage customer accounts, particularly because SendinBlue was dealing with multiple data sources and limited resources, Jeanroy says.
“Data science and machine learning are the future of many industries right now, and top among them is marketing technology,” Jeanroy adds. “We knew we needed to stay on the cutting edge of the marketplace by automating key internal processes, and we were also looking to eliminate or drastically minimize the margin of error inevitable with multiple parameters and human judgment. To continue to ensure that SendinBlue’s delivery system maintains the highest reputation and quality in this rapidly evolving space, basic and fixed rules were not efficient enough.”
To achieve that leap in efficiency, SendinBlue selected Dataiku Data Science Studio. The system automates checks that filter out potential spammers before they can access the SendinBlue platform.
“It combines multiple tools, including data manipulation across multiple technologies, machine learning capabilities, and an intuitive, user-friendly interface,” Jeanroy says.
Dataiku’s algorithms are based on multiple parameters, including historical data from more than 1 billion emails and associated events, including clicks, opens, and bounces; it also works off a list of thousands of blocked accounts and hundreds of fraud criteria, such as IP addresses, histories, sending volumes, and past behaviors. Using that information, Dataiku Data Science Studio analyzes new users and automatically classifies them as good, bad, or uncertain. Based on their risk scores, potential users can then be validated, blocked, or sent to customer care for further analysis.
The solution, which took about four months to fully implement, allows SendinBlue to scale by serving customers more quickly. This analytics initiative saved the equivalent of a full-time job.
On top of this, SendinBlue can now provide a better customer experience by considerably shortening validation delays. Initially handling just 24 percent of the new accounts as part of a staged rollout, the machine learning model now handles most of the new accounts. Thanks to those machine learning capabilities, SendinBlue can now validate more than 30 million email and text messages daily.
“The results have been tangible. Not only has the new solution helped improve our bottom line, but it has also decreased the amount of time previously spent by our teams on manual tasks,” Jeanroy says. “The process is now properly scalable to keep up with our rapid customer growth, and it’s also helped us improve SendinBlue’s customer experience by allowing for more control and feedback.”
Customers have benefited from an improved user experience, Jeanroy adds. “The automated, machine learning process allows for less waiting time, ultimately resulting in less frustration during the on-boarding process. It’s also allowed us to improve our response time to clients’ help requests, now that our customer care team spends less time moderating the validation process.”
Reduced fraud was another advantage of the solution.
“Dataiku helped us to massively scale our team’s productivity while delivering a better user experience,” adds SendinBlue’s CEO, Armand Thiberge. “We are now going to leverage the power of machine learning to aid our customers in real time while they use SendinBlue. As fraudsters’ systems always evolve, we’ll soon leverage a new generation of advanced algorithms to detect new kinds of fraud.”
The Payoff
Since implementing Dataiku Data Science Studio, SendinBlue has been able to:
- validate 30 million email and text messages daily from more than 50,000 companies;
- eliminate the equivalent of one full-time employee;
- dramatically speed up the on-boarding process for new customers; and
- reduce fraud.