The short answer: you need four things out of most invoices — who sent it, what they called it, when it was dated and what it comes to — and you need them in the same four columns for every supplier, no matter how differently each one lays its document out. Everything below is about getting from a folder of PDFs to that sheet.

Why invoices resist a spreadsheet

An invoice is a document written for a human to read and pay. There is no standard for where the total sits, what the supplier's reference is called, or which of the several dates on the page is the one your accounting system wants. Two suppliers can bill you for the same thing and print almost nothing in the same place.

That is why invoice data entry survives in businesses that have automated almost everything else. The work is not difficult; it is that every document needs a person to look at it and decide which number is which. Twenty invoices is an afternoon interrupted. Two hundred is a job.

The three problems that actually cost the time are worth naming, because they are the ones any method has to solve:

Layout varies per supplier. The label above the figure you want might be "Total", "Amount Due", "Balance Payable" or nothing at all.

The same field has several plausible candidates. An invoice often prints an invoice date, a tax point date, a delivery date and a due date. Picking the wrong one is not obvious until a payment run is late.

Tax is presented inconsistently. Some invoices show net, tax and gross. Some show only gross. Some break tax into several rates on one document.

Decide the columns before you touch the documents

This is the step people skip, and skipping it is why the first spreadsheet usually gets rebuilt.

Write down the columns your spreadsheet actually needs, in the order you want them, before you open a single PDF. A typical accounts-payable set looks like this:

Supplier
Invoice Number
Invoice Date
Due Date
Net
Tax
Total
Currency
PO Number

Two rules make the list durable. First, name each column after what it means to you, not after what any one supplier prints — you want one Total column, not a column called "Amount Due" that half your suppliers do not use. Second, include a column even if only some invoices carry it. A PO Number column that is blank for your utilities and filled for your subcontractors is useful; a PO number squeezed into a notes field is not.

If a value genuinely is not on the document, the cell should stay empty. An empty cell is a fact you can filter on. A guess in the same cell is a number that will be summed.

The manual method, done properly

If you are doing this by hand, do it in an order that lets you check yourself.

  1. Put every invoice for the period in one folder and name the files so they sort predictably — 2026-08-supplier-invoicenumber.pdf works well, because the file name then carries two of the columns.
  2. Build the sheet with the header row first, and freeze it. Add one extra column called Source file.
  3. Work through the folder in order. For each invoice, fill one row and put the file name in Source file. Never fill a column across several invoices at once; that is how a row slips.
  4. Type figures without currency symbols or thousands separators, and set the column format instead. A cell containing £1,240.00 as text will not add up.
  5. At the end, check two things: the row count equals the file count, and the sum of Net plus Tax equals the sum of Total. Those two checks catch most transcription errors between them.

The Source file column is the part worth keeping even when you stop doing this by hand. It is what turns "this figure looks wrong" into "open that file and look", which is the difference between a spreadsheet you trust and one you re-check.

When the PDF is a scan

Some invoices are not documents at all — they are photographs of documents, usually because they arrived by post and went through a scanner, or because someone sent a phone picture. A scan has no text layer, so nothing can read it by looking at the file's contents; it has to be recognised from the image first.

You can find out which kind you have in a few seconds. Extract Text from PDF opens a PDF in your browser and shows you the text it contains. If the invoice comes back as text, any method here will work on it. If it comes back empty, the file is a scan and the words on it exist only as pixels.

That distinction matters before you choose a tool, not after. A method that works beautifully on supplier PDFs generated from an accounting system will return nothing at all from a folder of scans, and the failure looks like a bug rather than what it is.

Copy and paste, and why it stops working

The obvious approach is to open the PDF, select the figures and paste them in. It works for one invoice.

It stops working for a stack for reasons that are worth understanding, because they apply to most quick fixes here. PDF has no concept of a table — a table in a PDF is lines drawn near some text, and the reading order the file records is often not the order you see. So a paste from a two-column invoice frequently arrives interleaved, and a paste from a table arrives as one column of values with no indication of which row each belonged to. Then you spend longer untangling it than typing would have taken.

There is also a subtler problem. Copying gets you the characters; it does not get you the decision about which characters were the total. That decision is the actual work, and it has to happen once per document however the characters arrive.

Getting the whole folder into one sheet

The method that scales is the one that inverts the loop: instead of opening each document and deciding what to take from it, you decide once what to take, and then run every document through that decision.

That is what ExtractToExcel does. You name the columns — the list you wrote down earlier — hand over the folder, and each invoice comes back as one row in an Excel workbook, with your column names as the header. The supplier's layout stops mattering, because you are no longer reading the layout; you are asking for a field.

The same shape works for anything you receive repeatedly, not only invoices. Multiple PDFs to one Excel describes the general version of the workflow, and how to consolidate multiple invoices into Excel covers what to do with the sheet once you have it — reconciling a month of supplier invoices rather than converting a single batch.

Two things to be aware of before you rely on it. A value that is not printed on the document cannot be extracted from it — no method can read a PO number off an invoice that does not quote one — so those cells come back empty rather than filled with something plausible. And a scan still has to be recognised from the image, so it is a different and less certain job than reading a text PDF; check first with the tool above.

Common questions

Which date should go in the Date column? The invoice date, meaning the date the supplier issued it, unless your accounting system posts on the tax point. Keep the due date as its own column rather than choosing between them — you will want both eventually, and adding a column later means revisiting every document.

What about invoices in other currencies? Read the currency from the document into its own column and leave conversion to the spreadsheet. An invoice that has been silently converted at some unrecorded rate cannot be reconciled against the payment that settles it.

How should credit notes be handled? As negative values in the same sheet, with a column that marks them. Keeping them in a separate sheet means every total has to be computed twice.

Can I get the individual line items rather than the invoice total? That is a different shape of problem — many rows per document instead of one — and it needs a different sheet layout to hold it. It is worth its own treatment.

Does the supplier's template need to be set up first? Not for the approach described here. You are naming the fields you want rather than mapping coordinates on a particular layout, so a supplier you have never billed with before needs no setup.

Before you build the sheet

Whatever method you choose, decide the columns first and keep the Source file column. Those two habits are what make the spreadsheet auditable, and they cost nothing at the point where they are cheap — before there are two hundred rows in it.

If your invoices are text PDFs and the columns you want are printed on them, the folder-to-sheet route is available today. If they are scans, find that out first, because it changes which methods are open to you.