Invoice PDFs into a spreadsheet

The job today

Invoices arrive as PDF attachments. Someone opens each one, finds the vendor, the date, the invoice number and the total, and types them into a sheet. Vendors lay their invoices out differently, so you learn where to look, and a new vendor means learning again. It's the same four fields every time, and it's easy to put a digit in the wrong place. A typo can sit in the sheet until a column won't add up.

What changes when it's automated

A small tool reads every PDF in a folder and writes one row per invoice into an Excel sheet. A review column marks which rows are fine and which need a person. If a field can't be read with certainty, the tool leaves it blank and marks the row. It doesn't guess.

The output is a normal Excel sheet, so you can sort it, filter it, or copy it wherever you keep your records. Your week changes from typing every invoice to checking the few that were marked.

The steps, in plain words

1. Choose the fields you want, usually vendor, date, number and total.

2. Send two or three sample invoices for each layout, as copies.

3. We build the tool and test it on your samples.

4. You run a folder of PDFs through it and check the marked rows.

What you can try on your own first

Open one invoice and try to select the text with your mouse. If you can, the PDF has real text, and a tool can read it. If you can't, it's a scan, and a person has to read it.

Then set up the sheet: one row per invoice, one column per field, in the same order every time. If you have Excel for Microsoft 365 on Windows: Data > Get Data > From File > From PDF. Older Excel versions and Excel for Mac don't have it. It can pull a table out of a PDF that has real text. Name your files the same way each time, so any invoice is easy to find.

Count how many different layouts you really receive. That number decides whether a tool is worth building.

When it's worth getting help

Help makes sense when the same few layouts keep arriving and the retyping never ends, or when the built-in import comes out messy. Know the limits first. The tool reads PDFs with real text, so scanned images are marked for a person. One document layout per job; a second layout is a separate quote.

Where to go next

We explain what we build and how we quote on the Documents to spreadsheet page. You can also see the invoice demo, built on made-up invoices, or read the common questions.