Invoice PDFs into Excel for bookkeeping, checked against the total

Convert an invoice now

Keep “Excel (.xlsx)” as the format. For a 1-page invoice “One per page” and “All on one” give the same result; for a 3-page invoice “All on one” puts every line item on 1 sheet, with an empty row where a page ended.

What comes out: a real test invoice

We built an invented invoice from Studio Lumen GmbH in Graz to Jane Cooper in Linz: a letterhead, a block with invoice number, date and customer number, a table of 4 line items (the fourth with a description that wraps onto a second line), and a totals block with net, VAT 20 % and total due, all in European number style. The result card said “Put 17 rows into a spreadsheet, with 15 numbers as numbers you can add up”. With the table’s column lines drawn, the line-item part of the sheet looked like this:

The invoice printed 1.286,50 and “EUR 1.543,80”; the sheet holds the numbers 1286.5 and 1543.8, and shows them in whatever format you give the column. The position numbers 01 to 04 stay text with their zero. The letterhead and the invoice number, date and customer number sit in rows 1 to 7, so nothing on the page is lost.

The same invoice without column lines converted with every amount right and in its row, but with 2 differences you should know about: the position number and the description shared column A (“01 Logo design, 3 drafts”), and the “Unit price” heading sat 1 column left of its prices. The cause was the invoice-number block above the table, whose text crosses the gaps between the table’s columns. The table on its own, without that block, came out in 5 clean columns. PDF to Excel wrong columns explains why, and the 2 formulas that split a shared cell.

Check it in 1 minute

1. Click an empty cell and type =SUM(E9:E12), the line amounts. On the test invoice it gives 1286.5.
2. Compare it with the Net row (=E14). Equal means every line arrived with the right amount.
3. Check the tax: =E14*0.2 gives 257.3, the VAT row. Then =E14+E15 gives 1543.8, the total due.
4. If a line has a quantity, check 1 product too: =C10*D10 must equal E10 (2 times 64.5 is 129).

If the sum is off, the difference is the size of the missing or misread amount, which points to the row. A wrapped description, like row 13, never breaks the sum: its amount stays in the row above.

The per-supplier routine

Invoices from the same supplier share a layout, so the work is front-loaded:

1. First invoice from a supplier: convert it and run the full check above. Note which columns hold the amount, the VAT and the total, and whether descriptions wrap.
2. Every invoice after that: convert, then compare only the sum with the printed net total. If the supplier changes its template, the sum stops matching and tells you.
3. Many invoices at once: drop them together on the converter. You get 1 .xlsx per invoice, named after the PDF.

Building the bookkeeping sheet

Keep 1 sheet for all invoices, with 1 row per invoice (or per line item if you track costs by item):

Copy the numbers from each converted file with Paste Special, Values, so the sheet does not keep links to other files. Put the invoice number in a column formatted as Text before you paste, so a number like 000815 keeps its zeros. Add a check column with =D2+E2-F2; it must be 0 on every row.

Dates come over as text, exactly as printed. To sort by date, a formula turns 09.10.2026 into a real date: =DATE(RIGHT(C2,4),MID(C2,4,2),LEFT(C2,2)). More on that in PDF to Excel number formats.

When the sheet has to go into accounting software as CSV, the next steps are in PDF to CSV for accounting.

If the invoice is a ZUGFeRD e-invoice

A ZUGFeRD invoice is a normal PDF with the invoice data embedded as XML: every amount, the VAT and the invoice number as fields, written by the supplier’s software. That data is exact, and converting the printed page is a reconstruction of it. Drop the PDF on Inspect to find out: a ZUGFeRD shows “Has 1 attached file” with a name such as factur-x.xml.

getPDF does not read that XML for you yet. Accounting software that supports ZUGFeRD imports it directly, which beats any conversion. If yours does not, converting the page as above still works, and the sum check still applies. What ZUGFeRD and XRechnung are, and the German deadlines, are in E-invoices and business PDFs.

The honest part

The converter works out columns from where the text sits. A plain invoice with a clear line-item table converts well; one with lines drawn between the columns converts best. Designed invoices with coloured boxes, 2 tables side by side or amounts placed beside the descriptions need a closer look, and the result card warns about merged cells every time. The converter does not know which number is the total: you tell Excel that with your formulas.

A photographed or scanned invoice has no text, only pixels, and the converter stops with a pointer to OCR. Run OCR PDF first, then convert; amounts come through, but read the sum check twice, because OCR can misread a digit (Scanned PDF to searchable has the details).

And if your supplier offers a CSV export or an e-invoice, that file beats a conversion, ours included. The neighbouring job, statements instead of invoices, is in Bank statement to Excel.