What data transformation is
Data transformation is the cleaning and reshaping between raw data and a chart: removing test rows, fixing types, joining orders to customers, turning a column per month into rows, adding a margin column. It is the "T" in ETL. In many tools it means exporting data into another system first. In Klyrova it is a list of steps on a table of your data model, folded into one SQL query, so nothing is copied out to do it.
The editor
Open Transform data from a report's editor or a data model's page. It shows the model's tables on the left, the first 200 rows of the one you pick in the middle, and its applied steps on the right. Click a column header to pick it, right-click for its menu, double-click to rename it, or change its type from the badge beside its name. Click any step to see the rows as they were after it; a step that can't apply is marked, with the reason in words.
Steps you can use
- Rows: filter by values or by conditions, sort, keep the top or bottom rows, remove duplicates, remove alternate rows.
- Columns: remove, keep, reorder, rename, change type, replace values, fill down or up, split by a delimiter or by characters, merge columns, trim and change case, extract parts, round numbers, take date parts.
- New columns: a formula on each row, a conditional column from rules, an index, a duplicate of a column.
- Reshape: group by with counts, sums, averages, minimums, maximums and distinct counts; pivot and unpivot.
- Combine: merge two queries (left, right, full, inner, and the rows of one with no match in the other) and append queries.
Undo and redo walk your changes. Apply saves each changed table to the data model, in the right order when one builds on another, and every report on the model is worked out again.
Always in your database
Each step wraps the one before it, so the whole query runs in your database (or in the imported copy on Pro and Teams), and nothing is processed in the browser. Values you type into a step are passed as literals that can't break out of the query, and the result still goes through Klyrova's read-only check. A change that would break a report already built on the table is refused, naming the visual and the column it reads.
Or ask for it
Klyrova's agent can build transformation steps too: "make an orders table with only paid orders, its store's region, and a year column", and it adds the steps and saves the table, which you can open and change.
Questions about data transformation
Do I need to know SQL to transform data?
No. Each step is a menu or a dialog. Klyrova writes the SQL, and you can view it from the editor.
Where do the transformations run?
In your database, as one query, or in the imported copy when the model is imported. Nothing is copied out to transform it.
Can I combine tables from two databases?
A merge combines tables of the same data source. Tables from different sources can still sit side by side in one report.
Can I add calculated columns?
Yes: a custom column with a formula on each row, or a conditional column built from rules.
What happens to reports when I change a step?
Klyrova checks every report built on the table. A change that would break one is refused with its name; otherwise the reports are worked out again.
Updated 14 November 2026
