# Importing data

**Settings, then Workspace, then Imports**, then **New import**. Syncek reads
`.csv`, `.xlsx` and `.xls`, up to five thousand rows per run. Each object offers a
template file to fill in.

## The four steps

1. **Upload file.** Pick the [object](/core-concepts/objects-and-records) the rows belong to.
2. **Map columns.** Match each column to an [attribute](/core-concepts/attributes). Anything unmapped is
   skipped.
3. **Review values.** Clean up whatever did not land correctly.
4. **Preview import.** The last look before anything is written.

## Choosing what a repeat run does

- **Insert mode**: add every row, update existing records only, or both.
- **Matching attribute**: the field compared to decide whether a row is somebody
  you already have. [Text, Email, Phone, Number and Link only](/core-concepts/attribute-types).
- **Empty cells**: clear the existing value, or leave it alone.

## Fixing a messy file on the way in

Each column can carry up to ten changes, applied as the file is read: find and
replace, upper and lower case, parse as a number, convert a date format, extract by
position or by regular expression, fill empty values, add text before or after.

## If it goes wrong

Every run is kept in the import history, and **you can undo one**. The undo removes
the records that run created.

Getting the same data back out is [exporting](/your-data/exporting).