Member Platform and Document Pipeline for a Statutory Health Insurer

The Challenge

A statutory health insurer with several million members set out to digitalise application and advisory processes that until then ran on paper, PDF forms and postal mail. Members were to submit documents online, and advisors to process them in a structured way.

The hard part was not the frontend. Every journey ends in a handover to the insurer’s core systems - and those accept only a precisely normalised XML structure. Its documentation contradicted itself in several places, and the reference files supplied as the specification differed per document type in both field count and naming.

The Approach

Settle the interface before building the interface. From the diverging reference files we derived one harmonised master structure and made it the binding source for both sides - frontend and backend take their field names from it. That removed a class of error which would otherwise have surfaced only at handover to the insurer.

Fixed and overwritable fields separated. Certain values identify a case to the core system - source ID, case ID, document class, address origin - and must never be overwritten from the frontend. Others are legitimately variable. That separation lives in the data model rather than in a convention everyone is expected to honour.

Whitelisting instead of guesswork. Recognised document types are mapped explicitly; everything else lands in a defined catch-all rather than travelling into the insurer with a guessed value. In this context a misclassified document costs more than an unclassified one.

A platform, not a page. Built as a Nuxt frontend on a Directus headless backend, extended with custom endpoints, hooks and backend modules for the advisory workspace. Editorial and business teams maintain content themselves; development effort falls where something genuinely new is being built.

The Results

Members submit documents digitally, and the data arrives in the core system normalised and without manual rework. The media break between application and case handling is gone.

Beyond that, what counts for the insurer is traceability: every case carries its origin, its channel and its document class - in the same vocabulary the core system uses. Operations and data storage are GDPR-compliant and hosted in Germany.