Get Data — ingestion & bindings | ATX AnalyticsTable Handbook

Get Data — ingestion & bindingsPin a table, import a file, or pull a multi-gigabyte Parquet straight from a bucket. Get Data is where data enters a project. A table or saved query is pinned into a sheet at the cursor, and the pinned region becomes a binding — something that can be refreshed, re-configured and scheduled, rather than a one-off paste. File imports behave the same way: the binding remembers the file and the sheet, so Refresh re-reads the source and picks up changes. Import paths accept parameters, so a dated file name loads the right day's file on every run.

What it does

Pin a table or saved query as a preview (instant sample) or static (full result, cancellable background load)
Workbook (.xlsx) import with a remembered file and sheet, re-read on every refresh
Text/CSV import with delimiter auto-detection, quoted fields, junk-row skipping, encoding choice and malformed rows reported by line number
JSON and NDJSON import — nested objects flattened into dotted columns, arrays of objects into linked child tables
Direct cloud ingestion from Google Cloud Storage or Azure Blob: CSV, TSV, JSON, Parquet, .xlsx or a .duckdb database lands in the project database without passing through the app window
Random data generation for test and demo projects; Disconnect freezes a pinned region into plain values permanently

How it works

Place the cursor, choose the source, and pick Preview or Static — that single switch decides what the region can later do.
A Query Monitor shows progress on large loads and can stop one that is taking too long.
File and cloud imports are themselves schedulable, so ingestion runs unattended alongside query refreshes.
Parameters resolve at run time inside paths and file names, which is what makes a dated daily file work without anyone editing anything.
Every file-import connection is mirrored as JSON into the project's Import folder and re-synced on change and on save.

Practices that keep it reliable

Reach for direct cloud ingestion for anything above a few hundred megabytes — routing it through the window is the slow path
Set the pinned-row limit and the large-load warning threshold in Settings → Grid to match the machines in the field
Name bindings for the business object they carry, not the query that built them

In a regulated environment

Import definitions are configuration only and never contain data, so the Import folder can be committed and reviewed freely.
Because bindings are declarative, an ingestion estate can be audited by reading the folder rather than by opening every workbook.

Screens

Get Data — data sources on the left, file formats on the right.
Insert Table — pin a table or saved query, as a live preview or a full static load.
Text/CSV import — delimiter detection, encoding, junk-row skipping and line-numbered error reporting.
Workbook import — the binding remembers file and sheet, so Refresh picks up the new version.
JSON import — nested objects flattened to dotted columns, arrays into linked child tables.
Cloud import — browse a bucket or container and load Parquet, CSV or a whole DuckDB file directly.

ATX Institute on LinkedIn

This site needs JavaScript for the full experience. Every page above is reachable without it.