DocTable
Extract table from PDF — AI OCR table extraction
Updated October 2026
DocTable pulls a table out of a digital or scanned PDF and returns real rows and columns in Excel, CSV or Word, not a wall of text. Your first 3 pages are free — scan and download, no card needed. Try your first page without an account — sign in only to download it. After that, $3 buys 20 pages (one-time, no subscription). It reads pages with no text layer, rebuilds true rows and columns across multi-page tables, and handles Arabic, Persian and mixed-script documents.
| Item | Description | Qty | Amount (USD) |
|---|---|---|---|
| 1 | Consulting, May | 10 | 1,200.00 |
| 2 | Design review | 4 | 480.00 |
| 3 | Hosting, annual | 1 | 240.00 |
How it works
- Upload the PDF containing the table.
- AI OCR detects the grid — rows, columns and headers.
- 3 free pages, download included — see the extracted table first.
- Export to Excel, CSV or Word — included in your 3 free pages.
Why DocTable
- Detects table structure, not just character positions
- Handles multi-page and multi-column tables
- Wrapped rows stay one row instead of splitting in two
- 3 free pages, download included; then $3 for 20 pages
"Extract" is a different job than "convert"
Converting a whole PDF and extracting one table out of it sound similar but fail differently. A page can mix a table with a paragraph of notes, a letterhead, and a signature block — a naive converter drags all of it into your spreadsheet. DocTable's AI OCR identifies which region of the page is actually the table before it exports, so the header row, data rows and columns arrive clean, and surrounding text does not leak into your first column.
Multi-page and multi-column tables
Reports and statements often continue a table across several pages, repeating the header each time. DocTable can process a multi-page upload and keep the table structure consistent across the batch instead of treating each page as an unrelated file. Multi-column layouts (two tables side by side on one page) are read as separate structures rather than merged into one garbled row.
Wrapped text and merged cells
A long description that wraps onto a second physical line is a common source of broken exports — naive tools split it into two rows. Because DocTable reads the page visually rather than line by line, a wrapped description is reassembled into the single row it actually belongs to.
Same product rules as every DocTable scan
Your first page needs no account; sign in (Google or a verified email) to download it. Processing is temporary — see Security & data processing. For document-specific versions of this same capability, see bank statement to Excel and invoice to Excel; for the general converter framing, see PDF to Excel and PDF to CSV.
For teams: bulk table extraction
Pulling one table at a time doesn't scale for a customs desk or reporting team processing dozens of PDFs a month. The Business pack is a bigger one-time top-up built for that: $99 for 2,800 pages — a lower per-page rate than the packs above — with an official PayPal receipt for expensing and a Data Processing Agreement available on request. Uploads are still grouped automatically, so a batch of documents comes back as one clean file, not one file per page.
Get the Business pack — $99 for 2,800 pagesOne-time credit, not a subscription — top up again whenever you need to. No contract, no setup fee.
Frequently asked questions
What kinds of tables can it extract?
Single and multi-column tables, tables that span several pages, and tables inside scanned or photographed PDFs.
Does it keep merged cells and multi-line rows intact?
Yes — the vision model reads structure, so a description that wraps onto a second line is kept as one row instead of splitting into two.
What formats can I export to?
Excel, CSV or Word from the same extracted table.
Is table extraction free?
Your first 3 pages are free — scan and download, no card needed. Your first page needs no account; sign in only to download it. After that, $3 buys 20 pages (one-time, no subscription).
How accurate is it on complex tables?
Strong on clear layouts, including multi-page and multi-column tables; weaker on very dense or low-resolution scans — always verify before financial or legal use.
How do I extract table data from an image?
The same way as from a PDF: the page is read visually rather than parsed for an existing text layer. Upload a photo, screenshot or scan and the detected table comes back as rows and columns to download as Excel or CSV. This is the case that matters most, because a photographed report or a scanned page contains no text to parse — a parser-based extractor returns nothing from it.
How do I extract tables from a scanned PDF?
Upload the scanned PDF to DocTable; each page is read as an image and the tables come back as rows and columns in Excel, CSV or JSON.
Can I extract data from PDF to Excel automatically?
Yes. A batch of up to 100 files can go in one upload, and the REST API and MCP server run the same extraction from code; see the API.
Can I extract data from PDF to Excel with Power Automate?
DocTable has no Power Automate connector. Anything that can send an HTTP request can call its REST API instead; see the API.
How do I extract a table from a PDF in Python?
Text-layer libraries work on digital PDFs. For scans, call DocTable's POST /v1/scan with any HTTP client; there is no SDK.
Related pages
Temporary processing — files are not archived. Always verify OCR output before financial or legal use. Default OCR model is the same for every language. See Privacy in the app.