Insights in customer needs with AI-infused analytics

A few years ago, the Cera cooperative initiated a strategic plan for 2025. In order to give this strategy concrete form, Cera engaged Cegeka to consolidate data scattered across various databases into a singular, uniform data platform. Subsequently, Cegeka enriched this data using artificial intelligence (AI) tools and developed together with Cera’s various businesses flexible dashboards atop the data platform. These AI-infused analytics provide Cera with crucial insights into the behavior and needs of its members.

AI-infused analytics provides Cera with deeper insights into its members  

Cera, with more than 400,000 members, is one of the largest cooperatives in Belgium. It is primarily known as the largest shareholder in the KBC Group, but the cooperative also offers its members benefits in the form of discounts on numerous products, services, and events. Additionally, it supports over 700 social projects annually and provides advice on cooperative enterprise. 

In 2019, Cera embarked on a new strategic plan to enhance its engagement with members and gain a better understanding of their needs. Data concerning its members was available, albeit dispersed across various databases. Cegeka undertook the task of integrating them all into a singular, unified data platform. “This already provided us with a data-driven viewpoint of our members,” explains Valérie Bracke, Marketing Manager. “It yielded some useful global insights, such as the number of benefits used by our members. However, we wanted to delve deeper into the data, for example to know whether the interests of our members correlated with our offerings.” 

“We wanted to know whether the interests of our members correlated with our offerings.” 

 

Valérie Bracke, Marketing Manager, Cera 

Jan Bergmans adds that Cegeka knew their data very well because they already built that data platform. “Thus, when we expressed our interest in doing more with our data, they promptly proposed a solid vision regarding the feasible options. After discussing the possibilities, this produced some use cases we wished to implement. In the next meeting, Cegeka already offered us a roadmap for this implementation.” 

AI-infused analytics  

The first step was to enrich Cera’s existing member database. “Cegeka gathered the descriptions of all our offerings from our website,” explains Jan Bergmans. “This was unstructured data, so they used a large language model to extract some structured data from it. We supplied a list of fifty domains of interest, and Cegeka asked the large language model to rate each offer’s description based on these domains of interest. These scores were then added to the database, enabling queries on whether a specific offer lies within a particular domain of interest.” 

The database was further enriched with all the interactions members had with each of these offerings on the website, such as which pages they visited and which offerings they purchased. “Thanks to this AI-infused analytics, we now possess rich data regarding the supply and demand for these fifty domains of interest,” explains Valérie Bracke. “Cegeka then developed a dashboard that helps visualize this supply and demand. On the supply side, we can view the primary domains of interest covered by our offerings, while on the demand side, we can view the domains of interest that are most popular with our members. We can also filter on various parameters to understand the interests of specific target groups.”

Through this Power BI dashboard, Cera can now identify areas where supply and demand don’t match. For instance, if members are significantly interested in a specific domain of interest that Cera’s offerings do not adequately cover, the cooperative could consider expanding offerings in that area.

The fact that there already was a data platform proved to be a tremendous advantage, as Jan Bergmans notes: “The most important success factor of any AI project is data quality. Thanks to our data platform, we knew that a lot of data cleaning had already been accomplished, giving us a head-start on this project.” 

Personalization and prediction 

Building on these insights, Cera now aims to personalize communication for its members based on their interests, says Valérie Bracke: “Now that our data has been enriched with these domains of interest, we can ensure start working on ensuring that we don’t send a sports-related offering to a member known to lack interest in sports.”

Next, Cera seeks to predict which members are most likely to leave. “This is critical for our strategic plan,” emphasizes Valérie Bracke. We’re interested in finding patterns among members who opt out, so we can predict member withdrawals based on their interactions with our offerings. When we predict that a member is on the verge of leaving, we can take action immediately.”

From the onset, the partnership between Cera and Cegeka was pragmatic, in close cooperation and flexible, says Jan Bergmans. “We collaborated almost weekly, and we were highly satisfied with the progress. Cegeka presented a clear roadmap and implemented it step by step. However, they were open for change requests and readily adjusted their approach when new insights emerged.” 

“Cegeka implemented a clear roadmap, but readily adjusted their approach when new insights emerged.” 

 

Jan Bergmans, Cera