<@U04S8PLJWMT> I used connection from API Builder ...
# ask-ai
y
@kapa.ai I used connection from API Builder to BigQuery, why the bigquery table store the data as json instead of columnar?
k
By design, Airbyte is an ELT tool and first loads each source record as a raw JSON blob into the destination, leaving the structure unchanged. [[Full sync ELT](https://airbyte.com/tutorials/full-data-synchronization#step-1)] With the newer Destinations V2 (which BigQuery uses), Airbyte still loads raw records as JSON, but now uses BigQuery’s native
JSON
type instead of
STRING
when possible. [[Destinations V2](https://docs.airbyte.com/release_notes/upgrading_to_destinations_v2)] In other words: - The initial load from the API Builder → BigQuery is intentionally stored as JSON (raw data). - Columnar, typed tables come from the typing & deduping / normalization step. If you only see JSON, either: - You are looking at the raw tables (often in
airbyte_internal
), or - Typing/deduping did not run or complete. The docs do not provide step‑by‑step UI instructions for API Builder specifically, but they clearly state that Airbyte favors ELT and raw JSON storage first, then transformation to columnar tables. [[UX ELT guideline](https://docs.airbyte.com/platform/1.8/connector-development/ux-handbook#output-data--schemas); [Full sync ELT](https://airbyte.com/tutorials/full-data-synchronization#step-1)]
y
can airbyte json output as NDJSON?