# Prepare your data

> Browse rows and fields, fix what Golden flags, transform columns from the column menu, split and reshape data, ask the Prepare chat for changes, and see what you can prepare on a live warehouse connection.

Prepare · 7 min read · Updated 2026-09-28

Prepare is the spreadsheet-style view of your dataset. You can see every column and row, profile fields, fix and transform values, and filter down to the slice you want.

![The Prepare experience. You can clean your data, add calculations, create a dictionary, change settings, and get your data ready for analysis.](https://app.goldenanalytics.com/api/help/public-image?object=5720490c-e5ec-4238-8da0-a2e7a933a49f%2F1789304207255-2026-09-13_08-55-31.gif)

## Open Prepare

In the analysis sidebar, click Prepare. If your dataset has multiple source tables (a multi-table cloud connection), each table appears as a sub-item - clicking one narrows the view to that table's fields. Two tabs sit at the top: Table View for the rows, and Data Dictionary for field metadata.

## The toolbar

- Add field (the + button) - Add new field writes a calculated field with your own formula; Suggest new fields asks Golden to propose some.
- Search - narrows the visible rows.
- Visual Query Builder - add filter rules, then apply or clear them.
- Profile Columns - computes stats for fields that don't have them yet.
- Save Dataset - saves the data as a shared dataset for your team to reuse.

## The column menu

Click a column header to open its menu, titled Transform followed by the column name. At the top, Suggested Fixes lists problems Golden spotted in that column, such as Fill empty values with the share of empty rows, or Trim whitespace. Click one to apply it. The rest of the menu depends on the column's type.

| Column | Menu items |
| --- | --- |
| Text | Smart Split..., Merge Columns..., Trim whitespace, Uppercase, Lowercase, Find & Replace... (or Replace value... when the column holds numbers stored as text) |
| Number | Number Format, Currency, Decimal Places, Display Units, Default Aggregation, Round to..., Replace value... |
| Date | Date Format, Default Date Level, Extract... (Year, Quarter, Month, Week, Day, Hour, Minute, or Second as a new calculated field) |
| Any column | Fill empty values, Convert to..., Hide from field lists, Delete field |
| Calculated field | Edit Calculation..., which shows the formula |

Convert to... offers Dimension, Text, Geographic, Measure, and Date, minus whatever the column already is. Dimension and Text are different requests: Dimension changes how the field is used and leaves the values alone, which keeps a long numeric ID intact, while Text rewrites the values as text.

Hide from field lists keeps the data for formulas but takes the field out of the lists you pick from. Delete field is unavailable while a calculated field still uses the column.

## Right-click a cell

Right-click any cell in the grid for Edit value, Copy cell, and Copy row, which copies the row with tabs between values. On imported data you also get Remove row and Filter duplicates.

## Suggested fixes and other AI tools

Open the Data Agent pane with the pane button in the top bar (Toggle Data Agent pane). Click Show AI tools to see these, and Hide AI tools to give the space back to the chat:

- Suggested fixes - Golden scans the dataset and lists fixes grouped as Splits & Extraction, Restructure, Nulls, Outliers & Duplicates, and Formatting & Type. Each has a button to act on it, such as Fix, Split, Reshape, or Dedupe.
- Improve field names - proposes clearer names for your columns.
- Suggest new fields - opens Suggest New Fields, a list of calculated fields that fit your data. Tick the ones you want, or Select all, then click Add fields.
- Filter duplicates - finds and removes repeated rows, on imported data.
- Quick actions - the full transformation menu, grouped as String Functions, Date Functions, Number Functions, Type Conversions, Logical Functions, and Table Shape.

## Ask the Prepare chat

The same pane has a chat box (Message your analyst…). Describe a change in plain words - trim a column, split a field, add a calculated field - and the chat applies it as a step you can see and undo.

You can also delete rows by a condition. Ask to remove rows where Status is Cancelled, or to keep only the rows whose Notes mention a word. Removing rows changes the imported copy of your data, so it is available on imported datasets only.

The chat handles hidden fields too. Ask which fields are hidden and it lists them; ask it to hide or unhide a field - or a whole set, such as every ID column - and it moves them out of or back into the field lists while keeping the data behind them. It can delete hidden fields as well, but first it lists the ones it will remove and shows a single button, labeled with the count (for example, "Delete these 4 fields"), so nothing is deleted until you click to confirm. As in the column menu, it won't delete a column that a calculated field still depends on.

If you ask for a chart or a dashboard, the Prepare chat hands the request off, since Prepare has no canvas of its own. A chart request - "show me a chart of weight over time" - is built on a new Analyze sheet, and a dashboard request offers a button that builds the board in Communicate.

## Split a column with Smart Split

Smart Split... on a text column opens Smart Split, which shows how many values it analyzed and any pattern it detected. Pick how to split:

- Natural Language - type what you want in Describe what you want to extract, such as "Extract honorific" or "Pull out state".
- Delimiter - split on a character. Common delimiters include comma, pipe, hyphen, slash, colon, tab, space, and semicolon.
- Regex - pull out matches of a pattern, with presets such as Extract numbers or Inside parentheses.
- JSON - turn each key of a JSON value into its own column.

Name the new columns if you like, then click Split.

## Reshape columns into rows

Data that arrives with one column per year, state, quarter, or date has its values hidden in the header row, which makes it impossible to group or chart by that thing. Reshape Columns to Rows fixes that.

1. In the Data Agent pane, open Quick actions and pick Reshape Columns to Rows under Table Shape.
2. In the Reshape columns into rows dialog, confirm the columns whose names are really values of one thing. Golden pre-selects the ones it detected.
3. Name the new dimension and value columns, then apply.

> **Note:** Golden also offers this on its own when it spots wide columns after a load, so you may see the suggestion before you go looking for the menu item.

## Split combined columns

When a column looks like it holds more than one field - a place and a code stuck together, say - Golden asks what to do after the load:

- Split automatically - adds the new columns now and hides the ones they came from. You can undo it from the history.
- Review it first - shows each split with example values, so you can rename the new columns, leave one out, or keep the originals visible.
- Keep them as they are - leaves the columns alone. You can still split later from the column menu.

A preview under the choices shows what you'll get. If you dismissed the question, a banner above the grid offers Split… until you close it.

## Recipes and step history

Every transformation you run is recorded as a step. Click History in the top bar to see the sequence, undo back to any point, or remove a single step with Remove this step on its row - the rest of the steps are replayed without it. Save as Recipe keeps the sequence so you can re-apply it to similar datasets later.

## Prepare a live connection

A live dataset stays in your warehouse, so Prepare can't rewrite its rows. You can still shape how Golden reads it:

- Convert to... Dimension, Measure, Date, or Geographic. Text isn't offered, because it would rewrite the values.
- Number and date formats, Default Aggregation, and Default Date Level.
- Calculated fields, including date parts from Extract... Golden turns the formula into your warehouse's own SQL when it runs.
- Smart Split with Natural Language or Delimiter. The parts become calculated fields.
- Renaming a field through the chat, which changes the name Golden shows rather than the column in your warehouse.

> **Note:** Editing values, removing rows, text rewrites such as Trim whitespace or Find & Replace, and Filter duplicates need imported data.

> **Tip:** Click Save Dataset in the toolbar to promote your prepared data into a shared dataset so teammates can re-use it.

### Related

- [Create a calculated field](https://goldenanalytics.com/docs/calculated-fields)
- [Reshape columns to rows](https://goldenanalytics.com/docs/reshape-columns-to-rows)
- [Import a data dictionary](https://goldenanalytics.com/docs/import-a-data-dictionary)
- [Import files](https://goldenanalytics.com/docs/import-files)

Source: https://goldenanalytics.com/docs/prepare