The three panes
| Pane | Shows |
|---|---|
| Flow (top) | The steps and how they connect |
| Profile (middle) | Each field with a distribution of its values |
| Data grid (bottom) | The actual rows |
The flow pane is the recipe. The profile pane is what makes Prep worth using.
The profile pane
Every field shows a histogram of its values — how many distinct ones, how they are distributed, how many nulls.
That is the difference from writing SQL blind. You see immediately that a status field has four values plus one row spelled differently, or that 12% of a column is null, without writing a query to find out.
Click any value to highlight every row containing it across all fields. That one interaction answers most data-quality questions faster than anything else.
Prep versus Desktop
Desktop analyses and visualises. Prep shapes data before Desktop sees it.
You would use Prep for work that has to happen repeatably and would be awkward in Desktop: unioning many files, pivoting wide data, cleaning inconsistent values, joining sources at different grains.
A flow ends in an output — a .hyper extract, a published data source, or a CSV — which Desktop then connects to.
How a flow runs
Steps connect left to right. Prep works on a sample while you build, so the profile pane stays responsive, then processes everything when you run the flow.
What you see while building may not include the awkward rows. A value that appears in the full data but not the sample will not show in the profile pane — so check the output, not just the preview.