Five ways, from simple to smart
- Excel's built-in import — works for clean tables in text-based PDFs.
- Online converters — quick, but you upload documents to someone else's server; don't use them for invoices and personal data.
- Power Automate and similar tools — automate repetitive steps, best when documents always look the same.
- Python scripts — flexible for a programmer, but fragile when layouts change.
- AI extraction — reads any layout like a person and returns the fields you need, even from scans and photos.
Excel's built-in PDF import, step by step
- In Excel, open Data → Get Data → From File → From PDF.
- Choose the file; Excel shows the tables it found.
- Pick the table and press Load — or Transform to clean it first.
- Next time, replace the file and press Refresh.
It is free if you already have Excel, but it only works with text-based PDFs and tables that keep the same structure. Scans and invoices from different suppliers need something smarter.
Scanned PDFs and photos
A scan is a picture of text, so it first needs text recognition (OCR). Basic OCR turns the picture into text but often breaks tables apart. AI extraction handles both steps: it reads the scan and returns a clean row per document — supplier, date, amounts — ready for Excel.
When AI is worth it
- Documents come from many suppliers or banks, each with its own layout.
- Some are scans or photos.
- Someone spends hours a week retyping.
- The data must land in a specific place — Google Sheets, a CRM, an accounting system.
Privacy: converters or your own tool
Invoices and statements contain names, account numbers and amounts. Free online converters process them on servers you know nothing about. A tool of your own can run on your server or cloud account, use AI providers' business APIs that don't train on your data, and keep a log of who processed what.
How much it costs
| Way | Cost | Good for |
|---|---|---|
| Excel import | Free with Excel | Clean tables in text PDFs |
| Online converters | Free or a small subscription | Occasional, non-sensitive files |
| Automation platforms | Platform licences | Identical documents in large numbers |
| Custom AI tool | from $500 + a few cents per page | Varied layouts, scans, data into your system |
More on pricing in how much AI document extraction costs.
Example: statements from five banks
A company receives monthly statements from five banks, each in its own format. The tool watches a mailbox, reads every new statement, converts all of them into one table with the same columns — date, counterparty, purpose, amount, balance — and adds the rows to a shared spreadsheet. The accountant checks only the flagged lines instead of retyping five statements.
Questions
Can it handle bank statements?
Yes — turning statements from different banks into one clean table is a typical task.
Can I convert a PDF table to Excel for free?
If the PDF is text-based and the table is clean, Excel's built-in import does it for free.
How do I convert a scanned PDF to Excel?
The scan needs text recognition first; AI extraction does recognition and table building in one step.
Can it run automatically for every new file?
Yes — the tool can watch a mailbox or folder and add new rows to your spreadsheet as files arrive.
Is my data safe?
The tool can run on your server, and I develop it on anonymised samples.