How Hedvig Scaled Self-Service Analytics and Maintained 2-Minute Claims with Google Looker

Hedvig is a Swedish insurance technology company built for tech-savvy customers, offering a mobile-first experience defined by its 2-minute claim turnaround — from filing to payment. That operational promise requires more than product speed; it demands that every team across the business reads from the same data. As Hedvig scaled from startup into a growing insurer, inconsistent metric definitions became a structural risk. Marketing, underwriting, and pricing teams were working from siloed data views, and the same business event — a sale, a claim, a renewal — meant different things depending on who was reporting it. In insurance, that kind of ambiguity isn't just inefficient, it's a compliance and actuarial liability.

The data team became a bottleneck. Ad hoc reporting requests consumed the bandwidth of data scientists who should have been building pricing models and risk tools. The team needed a way to give business users reliable, self-service access to trusted metrics without becoming a permanent support desk for dashboard requests.

Hedvig deployed Google Looker as its semantic layer, backed by BigQuery as the data warehouse and dbt for transformations. The combination created a single, governed definition for every key metric across the organization. Business teams in claims, underwriting, and sales could now build their own dashboards using Looker's interface, with confidence that their numbers matched those in every other department. Looker's LookML Liquid templating enabled a sophisticated capability: the same dashboard could dynamically switch between live operational data and frozen historical snapshots, giving actuaries and business managers the precise views they needed without maintaining duplicate reports. CI/CD integration in Looker caught modeling errors before they reached production.

The impact was measurable at both the team and business level. The data team shifted from reactive reporting to proactive model development — specifically predictive pricing models that directly support Hedvig's underwriting accuracy. Headcount expansion that would otherwise have been required to absorb reporting demand was avoided. The 2-minute claim turnaround remained intact, now supported by instant access to reliable operational data rather than manually compiled reports.

Hedvig is now refactoring its data models for an AI-driven future, making them more self-explainable and accessible to language model agents. The semantic layer built on Looker is positioned as the foundation for conversational analytics — enabling natural language queries about profitability and operational performance from non-technical staff. The same infrastructure that eliminated data silos in 2024 is being readied for the next generation of AI-augmented insurance operations.