How the converter works
The converter operates in three distinct steps.
The first is DTSX parsing. A .dtsx file is a proprietary Microsoft XML that describes an entire SSIS package: source and destination connections, Data Flow Tasks (the actual transformations), control flow (conditions, loops, error handling), and variables.
The second step is logic translation. Each SSIS component is mapped to its DBT/Snowflake equivalent. An OLE DB source becomes a reference to a Snowflake table (via a source model). A Derived Column becomes a SQL expression in a staging model. A Lookup becomes a JOIN.
The third step is code generation. The converter produces a complete DBT project: SQL models (staging, intermediate, marts), the schema.yml file with tests and documentation, and the dbt_project.yml configuration.
What the converter handles well
Standard cases represent 60-80% of a typical SSIS estate. These are packages that extract from one or more SQL sources, apply classic transformations (joins, filters, aggregations, type conversions), and load into destination tables.
The converter also handles T-SQL to Snowflake SQL translation: CTEs (Common Table Expressions), window functions (ROW_NUMBER, RANK, LAG/LEAD), data types, date and string functions — all converted automatically with functional equivalence testing.
What the converter doesn't handle (yet)
Complex cases require manual intervention. C# or VB.NET Script Tasks (embedded .NET code for non-standard logic) are not automatically convertible. Cursor-based T-SQL, dynamic SQL, and cross-database references require expert rework.
For these cases, the converter generates a commented skeleton indicating what the SSIS component did and suggesting a Snowflake approach. The senior consultant doesn’t start from scratch — they have a starting point that accelerates work by 2-3x.
Difference from SnowConvert: Snowflake’s official tool (SnowConvert AI) also offers SSIS → DBT conversion since late 2025. The key difference is that SnowConvert converts without auditing. It processes everything — including obsolete packages that should be decommissioned rather than migrated.
Observed results
On projects delivered with the converter, the measured gains are as follows. Conversion time is reduced by 60-70% compared to manual rewriting. The conversion error rate is below 5% on standard cases. Documentation is generated automatically.
The most unexpected benefit is documentation. Each generated DBT model includes an origin comment (“converted from SSIS package XYZ, Data Flow Task ABC”) and basic tests (not_null, unique on key columns). For the first time, the estate is documented and testable.
Learn more: discover our full methodology or view our case studies to see the converter in action on real projects.