Benodigde gegevens

Data Requirements for Using Onesurance.

Successful implementation of the Onesurance requires timely and complete access to relevant customer, policy and claims data. In this article, we describe the minimum data set required to train and deploy AI models such as Top Defend, Top Sales and Top Fit, and explain how this data is ideally shared.

1. Overview of required data tables.

Onesurance uses four core data sets. These can be delivered via database, file or API.

Below are the minimum columns required for each table:

Table: Relationship (customer level)

• Relationship ID

• Year of birth

• Zip code

• Region or department

• Segment (e.g., private / business)

• Relationship group or reference to main relationship (e.g., family/business)

• Origin of relationship (e.g., direct, through broker, acquisition)

Table: Policy (insurance).

• Policy ID

• Relationship ID

• Effective date / start date

• End date / expiration date

• Type of insurance or main industry (e.g., auto, liability)

• Type of product or industry (more specific)

• Annualized net premium

• AFD industry code (optional, for standardization)

• Status (ongoing, terminated, etc.)

• Reason for churn (optional)

Table: claim (s)

• claim ID

• Policy ID

• Relationship ID

• Damage date

• Settlement date

• Debt damage (yes/no)

• Claim expense (amount)

Table: Contact moment (interactions).

• Relationship ID

• Date contact time

• Type of contact moment (e.g., mail, phone, portal)

• Settlement date

• Employee (optional)

• Subject or reason for contact (optional)

2. Minimal vs. expanded data set

The above columns are the minimum basis needed to run initial models. If additional columns are available, such as customer characteristics, payment data or communication history, they can provide additional value in the analysis and model performance.

We are happy to think along with you about the maximum use of your data within the frameworks of AVG/GDPR and duty of care.

3. Data delivery options.

Onesurance supports a variety of methods for retrieving or receiving data. Our preference is for automated, structural links, but we are flexible on a situation-by-situation basis:

Preferred order:

1. Direct database linkage

   The customer or partner makes a database available (e.g., via read-only access).

   Onesurance periodically retrieves the data via a secure connection.
   
   

2. File drop on Azure Storage

   Client periodically provides files in a pre-agreed structure (CSV, Parquet).

   Onesurance processes them through an ingest pipeline in Microsoft Azure.
   
   

3. REST API

   The client provides a REST API from which Onesurance can retrieve the data.

   Suitable for partners with pre-existing data access through services.

All methods include strict security, logging and data validation. During the onboarding, we agree together which method best suits your situation.

4. Summary

Do you have questions about data delivery, mapping to our model or data completeness? If so, please contact your Customer Success Manager or the data integration team.