This audit played a pivotal role, providing a clear, strategic foundation for what needed to be retained, optimized, or transformed when it comes to DWH & BI. Building on the audit’s findings, BTA opted for a carefully staged DWH migration to Azure Synapse. Throughout the transition, teams navigated the complexity of running parallel environments and placed a strong emphasis on a controlled, coordinated shift, ensuring that the move to the cloud was, above all, a proactive step to future-proof BTA’s analytics capabilities.

BTA has been partnering with Scandic Fusion on data & analytics solutions since 2012. In the dynamic world of insurance, BTA has always understood the value of robust analytics and a sound data backbone. As new data sources emerged, organizational changes took place over time, smaller and larger improvements were made over time and business and data analytic needs shifted, it became clear that staying ahead meant more than just keeping up, it required intentional, forward-thinking change. Although their on-premise DWH was running smoothly and met best-practice standards, covering different aspects of data analytic needs simultaneously for three Baltic BTA companies, BTA and Scandic Fusion decided not to simply preserve the status quo. Instead, the aim was to proactively prepare for the best, setting the stage for a DWH migration from on-premise to cloud that aimed to build a foundation for future growth and resilience.

The Catalyst for Change

BTA’s long-standing on-premise Data Warehouse had grown into a stable powerhouse, supporting complex analytics for three Baltic companies and integrating evolving sources, AI solutions and business requirements. Still, BTA always thinking ahead, knew that staying at the top meant taking a closer look at how well their DWH & BI technology stack was well positioned for the future. That’s why, in 2023, thorough audit was kicked-off led by BTA & Scandic Fusion, digging deep into architecture, business and data analytical needs.

It became clear that while the existing setup was robust, a shift to a data warehouse cloud-based solution could better support BTA’s forward-looking strategy. Crucially, the audit provided an evidence-based foundation for this decision, demonstrating the value and necessity of migration, rather than only evolving the existing setup.

Every time a client insures a car or has doctor’s appointment and uses insurance policy, in insurance business it creates data. One event may not seem significant, but across thousands of customers, multiple insurance products, and countless interactions, the data quickly grows to a scale that requires a powerful data warehouse and analytics environment. A Proof-of-Concept validated Azure Synapse’s suitability to BTA’s data warehouse specific requirements, and the management and business approved a staged migration. The audit and PoC ensured every step were deliberate, risk-aware, and justified by measurable benefits. Decisions also were made related to business analytical platforms and more than one was selected.

Preparing for Migration

Instead of taking the high-stakes “big bang” migration route, BTA and Scandic Fusion agreed on an iterative migration approach for the “lift-and-shift” of on-premise DWH to Azure Synapse DWH. This wasn’t just about risk aversion, it was about ensuring that at every stage, business as usual could continue. Not to mention that this “lift-and-shift” included several hundreds of active database objects (data model with dimensions and facts) with varying degree of complexity excluding all technical supporting and configuration tables.  

By phasing the migration, the team could translate insights from each completed stage into smarter, smoother next steps, all while keeping data analytics and daily operations undisrupted to a minimum as possible. The gradual go-live approach was based on the readiness of each data area. At the same time, downstream data consumers such as Business Intelligence platforms had to adapt and operate with both data warehouse platforms in parallel.

• Migration Scoping and Cleanup: BTA used migration as an opportunity to carefully audit and rationalize existing data analytics assets, identifying unnecessary objects that would not be shifted to new DWH.
• Parallel System Operations: During migration, both the on-premise DWH and Azure Synapse data warehouses operated simultaneously. This dual-environment period, while complex, ensured uninterrupted data analytics and gave teams the ability to test, validate, and adjust without significantly affecting business data analytics and dependent processes.
• Incremental Data Loading: Special attention was given to minimizing data transfer and data loading times. Data from the core insurance system was transferred to loading all objects incrementally, optimizing data transfer between on-premise core insurance system to cloud.
• Automated and Manual Testing: Regression and hands-on testing, both automated and manual, became pillars of the transition, safeguarding data quality across old and new data warehouse platforms. This becomes more challenging once development initiatives are done only in the new environment; hence, diverging from the old environment that could be used as reference points for testing purposes.

Lift and Shift

A DWH database of a different architecture meant that the code could not be directly transferred 1:1, it had to be adapted to the "format" of the new database. This applies both to the code itself in terms of functionality as different/non-existent database functions, and also to important nuances that affect the way data is technically stored.

Data warehouse was not only migrated from on-premise to Synapse but also involved using different ELT tool – DBT. Although this tool has a lot of advantages, it has certain limitations that had to be considered when transferring the code and building the new load process. DBT is only a transformation tool, so all file loading, email sending, load orchestration, etc. had to be implemented elsewhere (via Azure pipeline or python scripts). DBT works at the table level. It was no longer possible to conveniently process (supplement) one table in numerous different steps, but had to look for ways and make logical changes to adapt to the tool's capabilities.

When transferring the old code, also some imperfections were discovered along the way that were also fixed during the lift and shift process towards the new DWH, resulting in more correct and consistent data. This also included that when transferring the code, various naming improvements were made along the way that all database objects would be named using a uniform improved approach to make it easier to search and interpret their meaning. On top of that the code was improved where possible (combined repeatedly used parts of the code in one place, removed unnecessary constructs, etc.) diverging from approach to bluntly just transferring everything as-is without thinking it through.

At project initiation, a cost-efficient Azure Synapse resource tier was chosen. As the solution evolved and data volumes grew, resources were gradually scaled to ensure optimal performance and efficient data processing. Below is depicted technical architecture with main concepts and components towards which the lift-and-shift was performed to. During the migration project the team also wanted to improve where possible  data loading, data quality, client reports, and code readability to make it even better than the old solution.

Lessons Learned

The migration of whole DWH platform to Azure Synapse lapsed from 2024 spanning into 2025. It meant that it is extended period for development and managing two live environments at the same time. This journey offered several valuable insights for organizations contemplating a similar path:

• Start with a Rigorous Audit: The up-front investment in a comprehensive audit provided the critical insights, pros and cons and confidence needed for a successful migration and avoided the pitfalls of unnecessary or untimed changes.
• Iterative Delivery Reduces Risk: Phased migration enabled the team to maintain service continuity, gather feedback, and refine methods as the project advanced.
• People and Communication Matter: The capability and engagement of BTA and Scandic Fusion’s teams, paired with clear stakeholder alignment, were instrumental to success.
• Parallel Operations Demand Coordination: Managing two live environments requires careful planning but delivers the assurance of business continuity and risk mitigation during transition. New development initiatives in the cloud data warehouse serve as a strong motivator for full adoption. Prioritizing these efforts accelerates the transition away from legacy systems.
• Challenges with “Lift & Shift” Approach: A direct “lift & shift” approach for complex logic is rarely straightforward. Expect to implement workarounds and redesign certain components to fit the new environment.
• Proper Decommissioning is Essential: The migration’s final phase, responsibly shutting down the on-premise DWH, required as much rigor as the move itself. A successful migration extends beyond the data warehouse itself. Downstream users and dependent systems must also adapt, requiring coordination and adjustments across the ecosystem. Careful coordination ensured that legacy dependencies were fully transitioned before final decommissioning.
• Post-Migration Performance Tuning: Post-migration tuning and optimization are essential to achieve optimal performance. Fine-tuning queries, indexing strategies, and resource allocation significantly improve efficiency.

Outcome: Prepare for the Best

By the close of 2025, BTA had successfully completed its phased journey to Azure Synapse, staying true to a business-first approach grounded in carefully managed change. The result: a robust, future-ready data warehouse built not just for today’s demands, but to support ongoing innovation. Every step – from defining what to migrate, to switching off the last on-premise DWH dependency – was shaped by clear purpose, consistent teamwork, and a shared drive for excellence.