Governance structures for data-based underwriting

Underwriting is increasingly based on internal and external data. Risk decisions, pricing and product design require consistent, accessible and compliant data. At the same time, regulatory requirements and operational complexity are increasing.

This article shows how underwriting departments can establish a functional data strategy and data governance structure to reduce risk, ensure compliance and increase operational efficiency.

An intelligent assistant also provides an in-depth insight into the topic and answers specific further questions.

Data dependency in underwriting continues to rise

Underwriting decisions are based on structured and unstructured data. In addition to internal portfolio and transaction data, specialist departments are increasingly using external sources such as scoring models, ESG key figures and geodata. Technical foundations are in place, but challenges remain:

• Data is distributed in data warehousesdata lakes, data marts and shadow processes and technologies.
• Responsibilities for quality, timeliness and data completeness are unclear.
• Departments evaluate data without assured origin or validity.

This fragmentation jeopardizes well-founded decisions and increases regulatory risks.

5 structural weaknesses hinder decisions

1. separate data sources

Departments access heterogeneous systems. Interface capability of existing systems is limited. Information is not consistent.

2. lack of competence

Without clear roles for data owners and stewards Data quality non-binding. Ownership for data products is not operationalized.

3. high manual effort

Inspection processes are carried out manually. Media disruptions and time losses hinder decisions, resulting in shadow processes.

4 Regulatory gaps

Regulations such as the GDPR or the European AI Act require transparent, auditable data processes. Without documentation, the risk of audits increases.

5. limited integration

Important external information is not systematically incorporated into decisions. Potential from market and environmental data remains untapped.

Clear questions lead to a functional data strategy

• What data quality do underwriters need in terms of tariff, risk and sector?
• Who takes ownership along the data value chain?
• Which data streams require regulatory evidence?
• How is compliant data use for automated decisions ensured?
• How is external data validated and integrated?
• Which systems provide consistent, audit-proof information?
• How do underwriters check data independently and efficiently?

5 concrete levers for functional governance

1. define and anchor roles

A governance model with a data owner, data steward and governance lead creates clear responsibilities.

2. anchoring quality criteria in the process

Underwriting-relevant features (e.g. location, rate logic, risk factor) are subject to standardized DQM routines. Automated validation and correction routines ensure that incorrect or incomplete entries are identified and corrected before they are included in decisions.

3. document data flows transparently

All automated decisions are based on traceable data streams and proof of origin - a prerequisite for a reliable single source of truth.

4. enable departments

Underwriters access validation rules, metadata and feedback mechanisms. They evaluate data on a professional basis.

5. systematically use external sources

Standardized interfaces and validations enable the integration of external data into the operational process.

The goal is efficient underwriting with clear governance

• Consolidated data structure - Secure access to up-to-date, linkable information.
• Responsibility structure - Departments know the relevance and origin of the data used.
• Transparency & documentation - Decisions are justifiable and verifiable.
• Regulatory compliance - Use complies with legal requirements.
• Operational scalability - Model transferable to new products and channels.
• Streamlining potential through governance - Elimination of redundant review steps and processes.

Concrete benefits for departmental underwriting

In the process

• Reduced effort thanks to a reliable data basis.
• Faster risk assessment of complex profiles.
• Clarity about permissible and relevant data for each business transaction.

Operational (data stewards)

• Clear test routines for data quality-relevant features.
• Transparent documentation of changes and approvals.
• Access to consolidated metadata and validation rules.
• Strengthening operational data responsibility along the process chain.

Strategic

• Reliable basis for AI-supported models.
• Increased ability to provide information to supervisors and auditors.
• Faster product and tariff development through reliable data usage.

Recommendations from the field

"Solutions only work when the problem is understood."

Governance initiatives are only effective if they respond to specific technical challenges. Technical measures such as tools or catalogs are only effective if they address operational weaknesses.

Recommendation

• The starting point is the risk and decision-making logic in underwriting.
• This results in the definition of the relevant data requirements.
• Governance is based on real use cases, not system logic.

Tools for target-oriented implementation

• Governance model - RACI matrix for business and IT.
• Quality criteria catalog - Per business segment (e.g. trade, industry, life).
• Questionnaire - Professional self-assessment on the use of data.
• System map - Overview of all relevant data sources.
• Regulatory check - GDPR, AI Act, EIOPA guidelines.
• Integration example - Systematically integrate external weather data.

Conclusion

Data-based decisions in underwriting require stable structures. Data governance is not a technical by-product, but an operational success factor. Implementation does not start with IT, but with the definition of technical problems. Only then can a sustainable, compliant and scalable data strategy be developed.

Sustainable governance forms the foundation - However, data processes in underwriting can only be implemented in a scalable and automated manner with a suitable technological infrastructure. In the follow-up article, we shed light on how architectural principles, system landscapes and interfaces enable data-driven decision-making at an operational level.