Transforming insurance data at scale

THE CHALLENGE

Moving away from a tangled web of integrations

As Spain's largest insurance group, Mapfre has a strong focus on digital innovation while retaining its focus on financial stability and trustworthiness. Moving away from an aging legacy IT landscape was clearly a logical move, but untangling the business from monolithic applications and a complex web of tightly coupled, point-to-point integrations was a critical challenge.

“We had legacy applications implementing core business functions such as issuing motor insurance policies, as well as satellite functions such as analytical systems,” explained Jésus Martin, Senior IT architect at Mapfre. “This involved point-to-point connections, and data dispersed across multiple databases and over 200 systems containing duplicated information.”

It was a structure that led to uncontrolled data duplication across a range of systems, creating fragmented information stores and preventing a unified view of customers or business entities. As a result, the company struggled to access key information on customers, policies, or claims.

“We developed a purpose to centralize our data consumption and offer a single source of truth instead of a fragmented and disconnected environment,” added Ignacio Cabrera, Solutions Analyst at Mapfre. “Our other goal was improved data governance and quality. We wanted to ensure that our data complies with key rules as defined in our transformation plan.”

Mapfre’s reliance on on-premises core systems was also resulting in scalability limitations, high maintenance costs, and significant technical debt, which hindered the insurer's ability to operate efficiently and deliver value to its customers quickly.

“This dependence on a tangled web of integrations and dispersed data was creating business and technological silos, and generating very high maintenance costs,” said Martin.

“We needed to create agility by decoupling our data querying from our core systems,” added Cabrera. “We wanted to attain scalability strategically, while also defining ownership of the data and responsibility for its lifecycle.”

The solution involved a comprehensive overhaul of Mapfre’s environment and the introduction of MongoDB.

OUR SOLUTION

Reducing silos while increasing flexibility and autonomy

Mapfre implemented a large-scale system transformation, migrating from an application-centric model to a platform-based ecosystem, comprising nine functional business domains supported by two technical platforms for engineering and integration.

“The aim was to create a more modular organization, reducing silos while increasing flexibility and autonomy,” Martin explained. “We also implemented key principles to leverage the benefits of cloud and managed services, such as MongoDB Atlas, and ensure coherent integration between our on-premises and cloud infrastructures.”

A key component of this architecture is the implementation of “Golden Records”—single, accurate sources of truth for business entities—stored in MongoDB Atlas on AWS.

“Many core security systems remain on-premises with legacy technologies like COBOL, while application servers and legacy relational databases are used for the large transactional cores and the cloud component,” said Martin. “We can leverage their greater scalability with well-protected front ends and a new, more scalable database, specifically MongoDB Atlas.”

The Golden Records solution employs a hybrid architecture that uses change-data-capture (CDC) and an event-driven design with Kafka and Flink to stream data from legacy on-premises cores to the cloud in real time.

“The Golden Record is not a static snapshot of data or any business entity. It’s a dynamic concept and approach that provides accuracy and consistency,” said Cabrera. “We can strategically use the MongoDB persistence solution to provide this Golden Record with the optimized and specialized query capabilities that will allow us to scale not only in volume but also in query modes.”

To ensure consistency and scalability across the company, Mapfre industrialized the deployment using reusable archetypes and standard infrastructure provided by its internal engineering platform.

“By choosing MongoDB Atlas to manage the Golden Record, we can guarantee that we will not only solve current problems, but also any future challenges and obstacles regarding the query models,” added Cabrera.

OUTCOME

A single, complete, and accurate view

The Golden Record transformation has enabled Mapfre to establish a single source of truth, bolstered by clear data ownership and governance, thereby eliminating its previous fragmentation issues.

Mapfre is, as a result, moving away from fragmented, scattered data to a single, complete, and accurate view of business entities such as clients, policies, or claims. This centralized data environment that MongoDB Atlas delivers has also improved consumption and eliminated uncontrolled duplication.

“The Golden Record is the single, complete, and accurate view of a business entity on a functional platform,” Martin explained. “It represents the source of truth for our functional platforms as they appear to the outside world.”

By using MongoDB’s flexible document model, Mapfre has achieved massive scalability and can handle dynamic, polymorphic entities without schema fragmentation. Specifically, the architecture successfully decouples data querying from core transactional systems, ensuring that high-volume read operations do not impact the performance of legacy cores.

“With MongoDB and the Golden Record, we can encapsulate our architectural decisions so they are replicated across each and every functional platform,” said Cabrera. “We can have entities that implement various behaviors without replicating collections, without having to model that model. It offers us massive scalability, and we now have very high capabilities.”

On a practical level, this new data environment has delivered operational efficiency, streamlining key processes and reducing costs, enabling faster time to market for customer innovations. From the customer’s perspective, the impact is transformative: instead of dealing with fragmented records across 200+ systems, they now experience a seamless, trustworthy relationship with Mapfre. Requests such as GDPR-driven data deletion are handled quickly and consistently, giving customers confidence that their privacy rights are respected. Interactions are based on a single, accurate profile, eliminating confusion from duplicated or mismatched information. Overall, business systems are stable, fast, and reliable, and Mapfre’s operations are simplified, agile, and future-ready.

“The Golden Record’s native storage approach gives us that flexibility because it’s a dynamic product subject to constant change,” concluded Cabrera. “MongoDB Atlas plays a key role in giving us that.”