Datasets
Editing Datasets
Shape the data — rename columns, add calculated fields, run transforms, and tune caching after the dataset exists.
Column overrides
Open the dataset to see its columns in a side panel. Each column has a label, type and visibility toggle. Common edits:
- Rename
cust_email_addrto Email - Hide internal IDs from public reports
- Reorder so the most important columns come first
- Change a number column's display format (currency, percent)
Calculated fields
Add a new column derived from existing ones — for example total = price * quantity or month = date_trunc('month', created_at). Calculated fields appear alongside real columns; widgets can chart or aggregate them like any other.
When to use calculated vs. transforms
Transforms
The transform pipeline runs afterthe query — useful when you can't (or don't want to) change the source SQL. Available steps:
- Rename — alias columns
- Cast — convert types (text → number, etc.)
- Derived — add a computed column
- Filter — keep only rows matching a condition
- Deduplicate — drop repeated rows on chosen keys
- Sort — order rows by one or more columns
- Limit — cap the row count
Execution & cache
Switch between Live and Cached at any time from the dataset header. For cached datasets you can:
- Adjust the refresh schedule
- Trigger an immediate refresh
- Pause refreshes (useful when the source DB is in maintenance)
Refresh history
Every refresh logs the start time, duration and row count. If a refresh fails (timeout, syntax error, connection drop) the failure is captured with the database's error message so you can fix it without re-running the query manually.
Failure alerts
On the Team plan you can subscribe an email or a webhook to dataset failures so the right person hears about it before a stakeholder notices a stale chart.