# Import files: what Golden does with them

> What happens when you upload CSV, Excel, JSON or Parquet: re-adding a file, combining files or sheets split by period, choosing lists in a JSON file, naming a headerless file's columns, the 2 million row limit, Excel files that can't be read, and recovering tables that didn't load.

Connect · 6 min read · Updated 2026-11-01

Connect your data covers the basics of uploading a file. This article is about what Golden does with the file once you've dropped it in, and the choices it asks you to make along the way.

## Supported file types

Golden imports .csv, .txt, .tsv, .xlsx, .json and .parquet files. Anything else is refused with a message naming the file type and suggesting CSV, TSV, JSON, Parquet or Excel.

## Add a file that's already here

If you add a file to a dataset that already has a table with the same name - usually because your data changed - Golden asks what to do instead of quietly adding a second copy. The dialog is titled with the table's name and "is already a table here".

- Replace the existing table - swaps the new file's rows in. Charts and joins built on the table keep working, as long as the columns still match.
- Add the rows to it - keeps what's there and stacks the new rows on, with a column recording which file each row came from.
- Add as a separate table - leaves the existing table alone and adds the file next to it under a new name.

> **Note:** Replace is disabled for a table that already has another table's rows stacked into it, because the stack is rebuilt from both parts every time the analysis opens. The dialog explains why; take the stack apart first, or pick one of the other options.

## Combine files or sheets split by period

Exports often arrive in pieces: one CSV per month, or one sheet per year in a workbook. When you upload several files, or a workbook, whose tables share the same columns, Golden offers to combine them in a dialog titled "Files that look like the same table" (or "Sheets that look like the same table").

1. Pick the set you want to combine and click Continue.
2. Untick any file or sheet you want to leave out.
3. If the file names carry dates, tick "Also add a Period date column" to get a real date column you can put on a time axis.
4. Click Combine. The result is one table, with a Source file (or Source sheet) column naming where each row came from.

Click Keep them separate (for files) or Choose sheets instead (for a workbook) to import them the usual way.

> **Tip:** To stack two tables that are already in your data model, use Combine tables in the Connect canvas toolbar, or right-click a table and choose Combine With Another Table. Pick the table to keep and the one whose rows go into it.

## Import JSON

A JSON file can be a plain list of records, newline-delimited records, or a document that wraps its records. Nested objects inside a record are spread into their own columns. You can upload several JSON files at once, and files with the same shape are offered for combining like CSVs.

When a file holds more than one list of records - say orders and customers side by side - a dialog asks "Which tables should we import?" and lists each one with its record and column counts. Each list you tick becomes its own table. With several files, your choice applies to all of them. Tables are linked automatically only where they share a key, so you may need to connect them yourself in the data model.

Golden also looks for columns whose cells hold JSON, like a Meta_Data column in a CSV export. It offers to split them in a dialog saying the column "looks like it holds more than one field", with three choices: Split automatically, Review it first, or keep it as it is. Splitting adds the new columns and hides the original, and the step can be undone from the data prep history.

## Very large files

An import keeps at most 2,000,000 rows. A very wide file is cut sooner, because the limit is really on the number of cells: a 400-column file keeps about 100,000 rows. When a file is cut, a warning names both counts and makes clear the rows kept are the first ones in the file, not a sample across it.

> **Tip:** If you need the whole file, filter or split it before uploading, or load it into a warehouse and connect to that instead.

## Excel files Golden can't open

Golden checks an Excel file before trying to read it, and tells you what to do when it can't.

- Password protected - remove the password in Excel (File > Info > Protect Workbook > Encrypt with Password, and clear the box), save a copy, and upload that.
- Older .xls format - open it in Excel and use File > Save As to save it as .xlsx or .csv.
- Too large - a single sheet holds more data than can be read in one piece. Save each sheet you need as its own CSV and upload those.

## How columns are read

- Formatted numbers in a CSV - values like $159,003.00, 68.8%, 1,234,567 or ($500.00) are recognized as numbers and become measures you can sum. Text that only starts with a symbol, and ID-style values like ZIP codes with leading zeros, stay as they are.
- Excel durations - a time-only column such as h:mm:ss imports as a number of seconds that still displays as a clock time, so you can sum or average it. Totals over a day keep counting hours.
- Merged banner rows - a title merged across the top of an Excel sheet is recognized, so the real header row underneath is used for column names.
- Files with no header row - a CSV whose first row is data rather than column names loads with placeholder names like column0. Golden reads the values and names the columns for you as the file loads, with no dialog to confirm, and a toast reports how many it named. The renames land as a single step in Prepare's data prep history, so you can undo them there if a name is wrong.

## When a table doesn't load

In a dataset with several tables, the fields pane lists any table that has no data on screen, so its charts don't just sit empty.

- Didn't load - a table whose data failed to download. Click Reload next to it to fetch just that table, without re-importing the dataset.
- Missing its data - a table whose data isn't in the analysis any more. Your saved sheets and joins still refer to it. To reconnect it, add it back under the same name (for example a file with that name and .csv) using Open Connect. If the data is gone for good, click Forget, then Forget for good? to drop its columns and joins from the analysis.

### Related

- [Connect your data](https://goldenanalytics.com/docs/connect-data)
- [Import a survey export](https://goldenanalytics.com/docs/survey-import)
- [Model several tables together](https://goldenanalytics.com/docs/multi-table-data-models)
- [Prepare your data](https://goldenanalytics.com/docs/prepare)
- [Troubleshooting](https://goldenanalytics.com/docs/troubleshooting)

Source: https://goldenanalytics.com/docs/import-files