Build a Pipeline

Define extraction, transformations and run frequency.

A pipeline is the contract between your raw data and your dashboards: read this source, apply these steps, on this cadence.

Anatomy of a pipeline

Field	Type	Description
name	string	Human-readable identifier, unique in the workspace.
description	string	What the pipeline is for. Shown on the detail page.
data_source	id	The source whose raw data is read.
transformations	array	Ordered steps applied to the extracted rows.
frequency	enum	hourly, daily, weekly or monthly.
run_time	time	When a scheduled run should start, in UTC.
is_enabled	boolean	When false the scheduler skips this pipeline.

1
Open Pipelines
Go to Pipelines and select Create Pipeline.
2
Describe it
Give it a name and a one-line description. Future-you will thank present-you.
3
Select the data source
Only sources that exist in the current workspace are listed.
4
Add transformations
Add steps in the order they should run. See Transformations for the full reference.
5
Set the schedule
Choose a frequency and run time. You can also leave it manual and trigger runs yourself.
6
Save, then run once
Run it immediately to confirm the output before you rely on the schedule.

Frequency

Field	Type	Description
hourly	string	Every hour on the hour (UTC).
daily	string	Once per day at run_time.
weekly	string	Same weekday and run_time each week.
monthly	string	Same day of month and run_time.

Editing a pipeline never rewrites history. Existing executions keep the transformations they ran with, and the next run uses the new definition.

Set is_enabled to false to pause a pipeline without deleting it. Scheduled runs stop; manual runs are still allowed.

Last reviewed October 2026.