If you've ever sat down with a stack of supplier invoices and a blank spreadsheet, you know how the afternoon disappears. Vendor name, ABN, line items, GST, total — typed in by hand, invoice after invoice, hoping you don't fat-finger a total and throw out your reconciliation.
There are a few ways to get PDF invoice data into Excel without doing it all manually. Here's what actually works, from slowest to fastest.
Method 1: Copy and paste (free, slow)
The default option. Open the PDF, select the text, paste it into Excel, then spend ten minutes fixing the formatting because the columns pasted as one long string instead of separate cells.
This works fine for one invoice. It falls apart at scale — and it doesn't work at all for scanned or photographed invoices, where there's no text layer to select in the first place.
Method 2: Excel's built-in "Data from Picture" / OCR tools
Excel and Google Sheets both have some OCR (optical character recognition) capability built in or via add-ons — you point it at an image and it reads the text into cells.
The catch: OCR reads text, not structure. It doesn’t know that "Hot water system install — $850.00" is a line item with a description and a price, or that the number under "TOTAL" is different from the number under "Subtotal." You still end up manually mapping which cell is which field, especially once a vendor’s invoice layout doesn’t match what the tool expects.
Method 3: Template-based invoice OCR tools
A step up from raw OCR: dedicated invoice-processing tools that let you define a template — "the ABN is always in the top-right corner," "the total is always the last row" — and then apply that template to future invoices from the same vendor.
This is a real improvement if you're processing high volumes from a small number of repeat vendors. The weakness is the same word that makes it work: template. The moment a vendor updates their invoice layout, adds a new field, or you get an invoice from someone new, the template breaks and you're back to manual entry — or building a new template from scratch.
Method 4: AI extraction (no templates, no layout rules)
The newer approach — and the one that avoids the template problem entirely — is using an AI model to read the invoice the way a person would, rather than matching it against a fixed layout. Instead of "the total is always in cell X," the model is asked "what’s the total on this invoice," and it finds it regardless of where it sits on the page.
Practically, this means:
- New vendors just work. No template to build first.
- Layout changes don't break anything. If a vendor redesigns their invoice, the model still finds the same fields.
- Scanned and photographed invoices are handled the same way as clean PDFs, since the model reads the image directly.
The trade-off is that no extraction method — AI or otherwise — is infallible, so it’s still worth a quick glance at the output before you file it away. A good tool will flag low-confidence fields for you rather than silently guessing.
Try it: Ledgr Invoice
We built Ledgr Invoice to do exactly this — upload a PDF, JPG, or PNG invoice, and get back a clean Excel or CSV file with vendor details, ABN, line items, GST, and totals already in the right columns.
A few things specific to Australian invoicing that we built in from the start:
- Recognises ABN and GST fields automatically, alongside international tax ID formats (EIN, VAT)
- Two export formats: a readable Document layout for filing, or a flat Database layout (with a separate line-items sheet) for importing straight into Xero, a spreadsheet, or a database
- Files are processed in memory and never stored on our servers — the invoice is read, the data extracted, and the file discarded
- 3 free extractions a day, no account or credit card required
How it works, step by step
- Go to the /demo page
- Upload a PDF or image invoice (or paste invoice text directly)
- Review the extracted fields — vendor, line items, totals, GST — and the confidence flag
- Download as Excel (Document or Database format) or CSV
That's it — no template setup, no account required for the free tier.
Which method should you actually use?
- One invoice, right now: copy-paste is fine.
- Same vendor, same layout, every time: a template tool can work, but budget time to rebuild the template when the layout changes.
- Mixed vendors, scanned invoices, or you just want it to work without fiddling: AI extraction is the least fragile option, and for light volume it’s free.
If you're doing this more than once a week, it's worth trying the free tier of an AI-based tool before committing to anything paid.
See how Ledgr Invoice handles your actual invoices — 3 free extractions a day, no sign-up required.
Try the live demo