In today’s dynamic risk environment and proliferating data in every walk of life, insurers are struggling to harness the data effectively. To address this data deluge and drive business value, a mindset of thinking of data as a product is a necessity.
This approach requires well-designed strategy to deliver data that consumers want.
For some companies, the key is internal democratization of data to drive synergy and reduce the inefficiencies that data silos cause. For others, the key is to establish data stewardship and accountability for regulatory and compliance needs and to build fit-for-purpose data products (e.g., pre-fill for underwriters, risk scoring and location Intelligence).
Where to begin?
Data products are built incrementally as a minimum viable product (MVP) that is accessible via application programming interface (API), continually enriched with domain specific data and intelligence, version-controlled and governed in a federated manner. For consumers, the data product shields them from the complexity of identifying, acquiring and processing domain-driven data sources into insights for decision making.
For example, a prospective home buyer has access to data on estimated property value based on criteria such as location, square footage and property type. But many first-time home buyers face financial surprises post-purchase. Auto and home insurance premiums may increase. So may replacement costs when they need repairs. Meanwhile, homeowners may find travel times increasing because of traffic congestion.
Insurers could harness their wealth of data and expose it to customers as a data product for “improving livability.”
See also: Achieving a 'Logical Data Fabric'
How to pursue this domain-driven journey?
To pursue this journey, insurers need to organize data by domains (location, policy, claims etc.) and align their MVP to a defined purpose (e.g., improving livability).
- Location is the critical data domain. It breaks down into granular data attributes such as basic information, primary modifiers, secondary modifiers, spatial and hazards. Completeness is a key.
- The domain must be mastered via machine-learning-based models for de-duplication, anomalies etc. Enriching the data with external sources enables accuracy and trustworthiness and provides a holistic view of location and risk characteristics.
- Location intelligence must be based on claims and additional data sets such as aerial imagery, weather, crime and hail and wind risks.
- Federated governance should be enabled through version controls, domain ownership and cataloguing to allow discovery through meta-data.
- The data and insights should be published through an API interface. Varied insights can be generated based on the context, such as replacement costs, safety score and protection gaps.
- Agent apps should leverage large language models and and agent-based modeling framework to enable knowledge management and reasoning capabilities for decision making.
To sustain and make a difference
Learn from failure – Many enterprises embark on modernization journeys as a technology initiative, resulting in limited business adoption and value. The strategy must revolve around business value and adopt an iterative approach, focusing on democratizing value through product features and consumption archetypes.
A data-driven culture focuses on multi-disciplinary data product teams with business stakeholders, domain-driven use cases, a data platform to deliver and democratize access to these data products. The work must be governed and measured by key performance indicators (KPIs) and incorporate consumer feedback.
