# Credit definition

One credit equals one page. That is the whole pricing unit. The only question is what counts
as a "page" for a file that doesn't have pages, like a spreadsheet or an email. This page
explains exactly how we count.

## Formats with real pages

For formats that are already paginated, a page is a page. Nothing to think about.

| Format | A page is |
|:--|:--|
| **PDF** | one page of the PDF |
| **PowerPoint** | one slide |

A 10-page PDF is 10 credits. A 20-slide deck is 20 credits. Scanned PDFs that we OCR are
counted the same way, with no premium.

## Formats without pages

Word documents, spreadsheets, CSVs, and emails have no fixed pages, so we define one by the
amount of text we extract. We total the extracted text and divide by the threshold below,
rounding up, with a minimum of one credit per file.

- **Word and email: every 3,000 characters (roughly 500 words) is one page.** That is about a
  page of writing, so a credit costs about the same as it would for the same content in a PDF.
- **Spreadsheets and CSVs: every 10,000 characters is one page.** A spreadsheet flattened to
  text holds far more characters than a page of writing, because a grid of numbers is not prose.
  We use a larger threshold so a data-heavy workbook is not billed as if it were hundreds of
  pages of reading.

| Format | A page is |
|:--|:--|
| **Word** | every 3,000 characters of extracted text |
| **Email** | every 3,000 characters of extracted text (most emails are one credit) |
| **Spreadsheet / CSV** | every 10,000 characters of extracted text, across all sheets |

So a short Word memo or a normal email is one credit. A long contract costs in proportion to
how much it holds, the same way a long PDF does. A large spreadsheet costs in proportion to its
data, on a more generous scale than prose.

> Spreadsheets are counted by their content, not by the number of tabs. A workbook with one
> large sheet can be several credits, and a workbook with many tiny tabs can be a single credit.
> Listed email attachments are not parsed, so they don't add to the count.

## Why we do it this way

- **Predictable.** The rule is the same for every file: bigger documents cost proportionally
  more, smaller ones cost less, and you see the exact credits used on every job.
- **Fair across formats.** The same amount of content costs the same whether it arrives as a
  PDF, a Word file, or a spreadsheet.
- **No surprises.** There are no per-feature add-ons or hidden multipliers. One page is one
  credit, every time.

After a parse finishes, the exact credits used are returned on the job as `credits_used` (see
the [API reference](/docs/api-reference)). For the list of formats we accept, see
[Supported formats](/docs/supported-formats).