At Sanlam, South African Financial Institution, AI Helps Attract, Retain More Customers

DataRobot AI Platform is a world-class machine learning platform. I highly recommend it for its technical prowess, transparency, deployment options, and customer support.

Modeling with Ease and Transparency

At Sanlam, one of Africa’s largest and oldest financial institutions, data analytics influence sales, improve client retention, help manage expenses, and support key strategic initiatives.

When Sanlam began applying AI to its analytics, open-source options felt cumbersome to navigate and lacked critical explainability for business stakeholders and compliance.

“With open-source software, optimizing a model could take anywhere from two weeks to a month,” said Shabbeer Omar, Head of Advanced Analytics (Business Intelligence) at Sanlam. “And we needed to be able to explain the dynamics underlying the model build and its predictions.”

Accessible AI for Actuaries and Data Scientists

After engaging with the DataRobot team and its AI Platform, Sanlam immediately saw the potential for end-to-end automation to expedite and expand its AI efforts for both data scientists and actuaries. Through a managed cloud environment, the company uses the platform’s best-in-class MLOps capabilities, including model performance monitoring.

“What blew us away was the ease with which the AI Platform makes data science and machine learning accessible to everyone,” Omar said. “The slick interface really streamlines the model-building process, where DataRobot builds models using multiple machine learning algorithms and automates the optimization process for each ML algorithm given the data. This automation of the technical model build allows our team to focus on solving the business problem at hand.”

Omar and the team also appreciate the flexibility of the platform’s multiple AutoML deployment options, including a JavaScript embedding approach as well as an API integration between Sanlam and DataRobot.

With MLOps, they can monitor models in production for data drift and can see the features driving each model much more easily. By understanding those underlying data points, Sanlam can deliver essential explainability to stakeholders.

“Understanding how various data points contribute to the outcomes of the model is so important for us to establish transparent processes in the business,” Omar said. “We can easily show that our decision-making is traced back to the data.”

Increasing Sales, Lowering Lapses

At Sanlam, sales performance depends on the quality of the leads it delivers to its customer-facing intermediaries. Using the DataRobot AI platform, they shifted lead qualification to be more data-driven. They also understand the features behind the leads, which influence customer retention and marketing efforts as well.

The performance of our data-driven campaigns far exceeds that of marketing campaigns where leads are based on intuition. We find clients with a higher propensity to purchase a given product having two to four times higher conversion rates relative to clients with lower propensities to purchase. This allows us to select leads more intelligently for campaigns.

For customer retention, they merge the company’s historical experience and the platform to create lapse propensity models. The earlier Sanlam can identify customers at risk, the sooner they can enact a range of interventions.

By spotting the factors that influence at-risk customers, Sanlam found that something as simple as a welcome message can positively influence customer lapse rates.

“We’ve seen improvement in lapse stats across our risk business,” Omar said.

Models One Month Sooner

The AI Platform transforms predictive analytics at Sanlam by simplifying the onboarding of new data scientists and empowering actuaries.

“We’re finding more actuaries in traditional spaces wanting to move into data science,” Omar said. “Since bringing in the platform, we’re exposing them more to the exciting world of advanced analytics.”

All users find they have shifted from manual coding to solving business problems, with a focus on understanding the variables that affect outcomes.

“Our role as data scientists or actuaries is not merely to write complex code or build dashboards,” Omar said. “The key role that we play is to focus on solving complex business problems through data—and that’s the most critical factor that DataRobot enables.”

Before, creating models took up to a month. Now, they find the same model creation process takes only a couple of hours.

From the start, the team at DataRobot has gone above and beyond in providing both technical and hands-on how-to help to guide Sanlam in approaching their industry-specific use cases, measuring success, and increasing adoption.