> AI agents: this is one page from Mammoth Analytics documentation. The index of all pages as Markdown is https://docs.mammoth.io/llms.txt. Append `.md` to any docs URL, or send `Accept: text/markdown`, to get Markdown.

# File upload

> Upload CSV, Excel, text, XML, archive, PDF, and image files from your computer to create datasets, with automatic schema detection.

Upload files from your computer to create datasets in Mammoth. Mammoth detects the structure and column types after you submit. You can upload a single file, several files, or a whole folder.

![New Dataset dialog with Upload selected, a drop area listing supported formats, and Upload file and Upload folder buttons](https://docs.mammoth.io/api/v1/images/20260408_152338_Dataset_creation_option.jpg)

## Supported file formats

Mammoth supports the following file types:

### Data files

These formats are parsed directly into structured datasets with automatic column and type detection:

| Format | Extension | Notes |
| --- | --- | --- |
| Plain text | `.txt` | Delimiter auto-detected |
| Comma-separated values | `.csv` | Most common data format |
| Tab-separated values | `.tsv` | Tabs as delimiters |
| Pipe-separated values | `.psv` | Pipes as delimiters |
| Microsoft Excel | `.xls`, `.xlsx` | Files with multiple sheets show an **Extract sheets** action after upload |
| XML | `.xml` | Converted to structured tabular data |

### Compressed files

Mammoth automatically extracts compressed files during upload. The extracted contents are processed as if uploaded directly:

| Format | Extension |
| --- | --- |
| ZIP archive | `.zip` |
| Bzip2 compressed | `.bz2` |
| Gzip compressed | `.gz` |
| Tar archive | `.tar` |
| 7-Zip archive | `.7z` |

### Documents and images

These formats are processed using AI-assisted extraction, converting unstructured content into structured datasets. They appear only when enabled for your workspace:

| Format | Extension |
| --- | --- |
| PDF | `.pdf` |
| TIFF image | `.tiff`, `.tif` |
| JPEG image | `.jpeg`, `.jpg` |
| PNG image | `.png` |
| HEIC image | `.heic`, `.heif` |
| WebP image | `.webp` |

> CSV files are processed fastest and most reliably. When exporting from other systems, choose CSV over Excel when possible for better performance.

## How to upload files

### Upload a single file

This is the most common workflow for ad-hoc analysis:

1. Click **+ New Dataset** in your Data Library
2. Select **Upload** (this is the default option)
3. Either drag your file into the upload area OR click **"Upload file"**
4. Click **Submit**

The dialog closes right away and the file appears in the Data Library as a pending item. Mammoth then reads the file, scans it for security risks, and creates the dataset. A progress header shows the upload count ("Uploading N items", then "Upload complete"), and you can cancel uploads that are still running.

If you select several files individually, each file becomes its own dataset.

![New Dataset dialog with sales_data_messy.csv chosen in the Upload tab and the Submit button enabled](https://docs.mammoth.io/api/v1/images/20260408_152354_upload_file.jpg)

### After upload: what you will see

Open the new dataset from the Data Library to see the Dataset Detail Panel. It summarizes what was ingested and lets you configure how future data additions are handled.

The panel contains these sections:

- **Overview**—Views associated with this dataset, plus a preview of a sample of your data
- **Settings**—Controls for Update method, Schema handling, and Cell editing for future batches
- **Batches**—Version history of all data uploaded to this dataset
- **Automations**—Automations that use this dataset

See [Dataset Detail Panel](https://docs.mammoth.io/learn/data-library/dataset-panel/) for details.

### When Mammoth needs your input

If Mammoth can't finish creating the dataset on its own, the dataset's panel shows an **Action needed** section. Which action appears depends on the file:

- **Define Date Format**—Mammoth can't tell whether dates are US or UK format. See [Upload validation prompts](#upload-validation-prompts).
- **Review & Fix**—The file has more than one plausible way to be read. The review dialog shows how Mammoth read the file and has an instruction box where you describe how it should be read. Choose **Looks right** to accept, or **Import as is** to skip the interpretation. On the save step you name the dataset and choose whether to refresh it manually or automatically by watching a folder, then click **Create dataset** or **Create dataset + automation**.
- **Extract sheets**—An Excel file has multiple sheets. Pick the sheets or tables to import, optionally tell Mammoth how to read each sheet, and click **Save N datasets**. You can also choose **Import as is**.
- **What to import** (PDF)—Choose **All content**, **Only tables**, or a custom instruction describing what to extract.
- **Password-protected files**—Mammoth asks for the password before it can read the file.
- **Corrupt files**—The panel shows "This file seems to be corrupt" and lets you delete the file.

If the virus scan flags a file, it is rejected as a dangerous file and isn't imported.

## Upload a folder

When you have multiple related files to upload at once, use the folder upload option instead of selecting files individually.

### Steps to upload a folder

1. Click **+ New Dataset** → **Upload**
2. Click **Upload folder** and select the folder from your system
3. Click **Submit**

A folder becomes one dataset, with all its files combined.

### Folder upload requirements

- The whole folder must be under 50 MB, because it is uploaded as a single request
- Files should share a consistent schema (matching column names and types)
- Only supported file types can be added

**Best for:** Monthly batch files, multi-location data consolidation, large grouped exports

![New Dataset dialog after Upload folder, listing several date test files from a Dates folder with Submit enabled](https://docs.mammoth.io/api/v1/images/20260408_152410_Upload_multiple_files_from_folder.jpg)

## Advanced upload features

## Upload validation prompts

If your file has issues, Mammoth pauses processing and provides specific guidance directly in the Data Library. Common scenarios include:

- **Date format ambiguity**—Mammoth cannot determine US (MM/DD/YYYY) or UK (DD/MM/YYYY) format. The **Define Date Format** table in the **Action needed** section lists each affected column with a sample value and a format dropdown. Choose the formats, optionally select **Define date format for this project**, and click **Validate & Apply**.

![Define Date Format panel warning that Mammoth cannot tell US or UK dates, with a US or UK format list](https://docs.mammoth.io/api/v1/images/20260605_100332_Dateformatambiguity1.jpg)

- **Column mismatches**—Rows in the new file have fewer or extra columns than the existing dataset schema. Mammoth flags each invalid row with a specific error (e.g. "2 missing columns","3 extra columns") and lets you discard them individually or all at once.
- **Delimiter and header row**—If columns appear merged or column names appear as data values, open the Data Preview modal (**See all details**, or **Data settings** in the Settings section of the dataset panel). There you can set the delimiter, the column header row, the number of initial rows to skip, and the date format.

![Data Preview for office_main_unstructured.csv showing 15 invalid rows with 1 missing column errors and Discard all](https://docs.mammoth.io/api/v1/images/20260408_152428_DataPreviewmodal_with_missing_col_error.jpg)

## File size and performance

### Size limits

- **Upload size**: 50 MB maximum per upload request, so a folder counts as one upload of 50 MB in total (see [limits and quotas](https://docs.mammoth.io/learn/limits-and-quotas))
- **Files per upload**: Up to 1,000

For files larger than 50 MB, use a [connector](https://docs.mammoth.io/learn/connectors/) to your database or storage instead, or split the file.

## When to use file upload

**Best for:**

- **Ad-hoc analysis** - Quick exploration of vendor files
- **Historical data** - Loading archives for trend analysis
- **External data** - Partner-provided files, government data
- **Initial exploration** - Testing before setting up Live Connections
- **Small teams** - Shared files via email or cloud storage

**Consider alternatives when:**

- **Daily updates** needed - Use Live Connections with Dataset Refresh
- **Uploads over 50 MB** - Use a connector
- **Real-time requirements** - Use Live Connections
- **Complex automation** - Use Data Consolidation or a connector

## Troubleshoot file upload

### Upload fails or times out

**Problem**: "Upload failed" error or loading spinner never completes

**Solutions**:

- Check the upload is under 50 MB in total
- If the file is flagged as a dangerous file, it failed the security scan and can't be imported
- Verify file format is supported (see format list above)
- Try downloading and re-uploading file (may be corrupted)
- Check internet connection stability
- Contact support if file is known-good format under size limit

### Columns not detected correctly

**Problem**: Column headers missing or wrong, data appears in single column

**Solutions**:

- Ensure first row contains column headers
- Check file encoding (save as UTF-8 if special characters)
- Verify delimiter is consistent (comma, tab, pipe, etc.)
- Look for embedded commas in CSV (should be quoted: "Company, Inc.")

### Wrong data types assigned

**Problem**: Numbers show as text, dates as text, text as numbers

**Solutions**:

- Preview sample data to verify detection
- Use **Define Date Format** or the Data Preview modal for date and delimiter problems
- You can change data types later using the Convert Column Type task
- Common fixes: Account numbers (Number → Text), Dates (Text → Date), ZIP codes (Number → Text)

### Multiple files create multiple datasets

**Problem**: You selected several files and got one dataset per file, but you wanted one combined dataset

**Solutions**:

- Use the "Upload folder" button instead of "Upload file" to combine files into one dataset
- Ensure all files have identical column structure
- Check that file formats are consistent (all CSV or all Excel)
- Verify files aren't corrupted or empty

## File upload best practices

### File preparation

- **Standardize templates** for recurring uploads
- **Include column headers** in first row
- **Use consistent data formats** (dates, numbers, text)
- **Remove empty rows** at top or bottom of files
- **Save as UTF-8** if file contains special characters

### Naming conventions

- **Dataset names**: Descriptive and searchable ("Q4 Sales Reports", not "Data")
- **File names**: Include dates or versions for batch tracking
- **Column names**: Clear and consistent (`Customer_Name`, not `Cust`)

### Data quality

- **Validate before upload** - Check for obvious errors
- **Use consistent values** - "Yes/No" vs "Y/N" vs "True/False"
- **Handle null values** - Decide on blanks, "NULL", "N/A", etc.
- **Preserve data types** - Format numbers and dates consistently

> **Tip:** Create a "staging" project for testing file uploads before processing in your main project. This lets you validate structure detection and data quality without affecting production datasets.

## Next steps

After uploading your files:

1. **Explore your data** using [Explore Cards](https://docs.mammoth.io/learn/explore-discover/explore-cards-overview) to understand distributions and quality
2. **Transform if needed** using [Visual Pipelines](https://docs.mammoth.io/learn/transform/) to clean, join, or reshape data
3. **Automate recurring uploads** with [Data Consolidation](https://docs.mammoth.io/learn/automations/data-consolidation/) automation
4. **Create dashboards** with [AI-powered visualization](https://docs.mammoth.io/learn/dashboard/creating-dashboards/)
5. **Export results** to your [preferred format](https://docs.mammoth.io/learn/export-share/export-formats) or destination

---
Source: https://docs.mammoth.io/learn/data-sources/file-upload.md · Updated: 2026-10-13