Cutting a week-long benefits journey to 2 hours — across 26 funds.
The problem
Swedish & German unemployment benefits weren't broken because the technology was bad. They were broken because 26 different A-Kassa funds — each with its own data contracts, eligibility rules, internal politics and operational pace — were trying to serve one citizen through 26 different paper-heavy processes. Caseworkers had to re-key the same information into incompatible systems. A typical first assessment took five working days end-to-end. People who had just lost their jobs were waiting a week to know if they qualified for support.
What I owned
I was embedded as the Product Manager / Business Architect across the platform programme: discovery through phased rollout. That meant requirements discovery with caseworkers across multiple funds, defining the PRD and roadmap, prioritising backlog with the engineering org, and running the alignment workshops that turned 26 different stakeholder positions into one schema everyone could ship against.
Approach
The instinct most people had was to build 26 integrations. I argued the opposite: one canonical REST schema, with a per-fund migration runway. Each fund kept its existing internal system; the schema was the contract between them and the unified assessment platform. Funds with simpler workflows went first and proved the model; the more complex ones came later with confidence baked in.
The hard work wasn't technical. It was running two rounds of discovery workshops with caseworkers across multiple funds, then mapping each fund's actual day-to-day workflow against the proposed schema and surfacing every edge case before we shipped. Decisions were documented as auditable trade-offs — what we explicitly chose to support, what we explicitly de-scoped, and what was reversible in v2.
Outcome
- 96% journey-time reduction — first assessment went from ~1 week to 2–3 hours.
- 500K+ users served per year on the unified platform.
- 26 A-Kassa funds aligned on a single roadmap and one REST schema.
- 60–70% less manual casework on the back end — caseworker time freed for genuinely complex cases.
- The schema became the platform's long-term data contract for downstream services beyond the original scope.
What I'd carry forward
Two lessons. First — in regulated, multi-stakeholder environments, the value of a documented trade-off compounds. Every reversible decision needed an explicit rationale, and that rationale paid for itself the third time a steering committee challenged it. Second — discovery is not a phase, it's a posture. We kept caseworker interviews running through rollout, and most of the post-launch fixes came from the same workshops we'd used at discovery.