Skip to main content
Help & Documentation

Reshape columns into rows

When column names are really values - years, states, dates, survey questions - reshape them into rows so you can group and chart by them.

4 min readUpdated August 27, 2026

Plenty of real spreadsheets put values in the column headers - a column per year, a column per state, a column per survey question. Golden treats each of those as a separate measure, which means you cannot group or chart by the thing the headers actually represent. Reshaping turns those columns into rows and fixes that.

Run a reshape

  1. Open the Prepare view of your dataset.
  2. Open the transformation menu and find the "Table Shape" group.
  3. Choose "Reshape Columns to Rows" - its description reads "Turn year/state/date columns into rows".
  4. In the dialog titled "Reshape columns into rows", check which columns are selected and what the new columns will be called.
  5. Click Reshape.

What the dialog asks for

  • New column for the names - where the old column headers go. Golden pre-fills this from what it detected, so a run over year columns suggests a year-shaped name.
  • New column for the values - where the numbers go. Defaults to Value.
  • Split one column into separate measures - keep everything in one values column (the default), or give each value of a chosen column its own column so those measures stay chartable separately.
  • An "After" band previews the result, along with the projected row count.
Note: If no sample rows are loaded you will see "No sample rows loaded, so there's nothing to preview - the reshape will still apply to the full table." The transformation runs against the whole table either way.

Golden offers it when it notices

You do not have to spot the problem yourself. Prepare shows a banner when a run of columns looks like values rather than measures, naming what it saw - year values, state values, date values, period values, or responses to one survey question - with a "Reshape..." button beside it.

Discover raises the same thing before it builds anything, in a card headed with the count of columns it found. There you get three choices: "Reshape automatically", which applies it and re-runs Discover on the new shape; "Review it first", which opens the reshape in Prepare so you can change the selection and the new column names; and "Keep it as it is".

The two-pane picker for surveys

Survey tables usually want most columns reshaped and a handful kept. For those, the reshape shows two panes - "Become rows" and "Stay as columns" - with a filter box over each and buttons to move a selection between them. Columns you keep are tagged as Identifier or Demographic so you can tell respondent keys apart from the things you want to segment by.

Tip: A reshape is a recorded step like any other transformation, so you can undo it from the data prep history if the new shape is not what you wanted.
Heads up: The Table Shape group is filtered by connection type. Live SQL connections that query your warehouse in place do not offer reshape - import the data first.