Transformations everyone can see and trust.
Cleaning data usually means a tangle of nested formulas, hidden steps and constantly checking that no data is lost. Clean replaces that with a visible flow. You pull a raw table from your datahub, and every step you apply to it is laid out in the open, in order, for anyone on the team to read. Clean gives an overview of the data you are cleaning.
What it does
Clean is the transformation layer between raw data and the numbers you actually analyze. You take a table that Extract landed ( say raw.economic) and shape it into something analysis-ready by filtering, typed, joined to other tables, and reshaped.
Each operation is a step in a visible flow, not a formula buried in a cell. Filter, merge, pivot, join and every step is readable on its own with a table change overview, and the whole sequence shows exactly how the raw table became the final one. No M code to decipher, no manual checking against the source and no running around in the spreadsheet to see what cells are joined or used.
How a transformation runs
- Pull — start from a versioned table in your datahub, not a copy on someone's desktop.
- Build — add steps one at a time: filter rows, join raw.economic to another table, pivot, reshape. Each step is a node you can see and edit.
- Trace — full column lineage runs through every step, so you can follow any value back to where it came from and see exactly what happened to it along the way.
- Rerun — point the same flow at new data and run it again, on a schedule or on demand. The logic is reusable across projects and teammates.
The result is a transformation anyone can read, reuse, and easily verify . The lineage that started at ingestion continues here, unbroken, all the way to the number you analyse.