What is ELT (extract, load, transform)?
ELT (extract, load, transform) is a data-integration pattern where raw data is extracted from sources and loaded into a target store first, then transformed in place using that store's own compute. It reverses the older ETL order so you shape data after loading rather than before.
ELT emerged with cloud data warehouses that made storage cheap and compute elastic. Instead of transforming data in a separate staging engine before it lands, ELT loads raw data first and runs transformations inside the warehouse.
- Extract pulls records from source systems (databases, apps, files).
- Load drops the raw data into the destination, untouched.
- Transform reshapes, cleans and joins it later using SQL in the warehouse.
Teams like ELT because raw data is preserved, transformations are version-controlled, and you can re-run them without re-extracting. The trade-off is that you still stand up and maintain a warehouse plus a transformation layer, and every question depends on those models existing first.
iDBQuery takes a different route for answering questions: rather than extract, load and transform your data into a warehouse before anyone can query it, iDBQuery connects to your sources and builds one live queryable model across them, then writes the SQL on demand and cites each figure back to its exact source row. You get an answer in seconds without building or maintaining an ELT stack, and if you already run ELT into a warehouse, iDBQuery simply connects to that warehouse as a standard SQL source.
Updated 2026-06-22