Turn a scanned invoice bundle into a VAT-ready schedule.

A single PDF holding two hundred invoices, running together with no separators and no text layer. InvoicePro works out where each one starts, reads it, checks the arithmetic, and hands back a spreadsheet with the doubtful rows already set aside.

Priced per invoice · Receipts and non-invoice documents not charged

The moment this is built for

A year of purchase invoices Year-end accounts prep Quarterly VAT review Catching up a client's backlog Due diligence Disclosure bundles
How it works

A merged scan to Excel in four steps.

01

Upload the bundle whole

You don't split it first, and you don't tell it how many invoices are in there. Most real scans have no text layer at all, which is the case it was built for.

02

It finds the boundaries

It works out where one invoice ends and the next begins. On a real scanned bundle it found 12 documents out of 12, with none missed and none invented — we check for both, since a tool that splits too eagerly would still find every real boundary.

03

It reads and checks each one

It reads the date, due date, supplier, reference, description, net, VAT and gross, then checks that net plus VAT equals the gross to the penny, that the VAT rate is 20%, 5% or 0%, and that any line items sum to the header. Scanned documents go on their own sheet, so you always know which rows came off a scan.

04

You check its working

Each row is shown beside the page it came from, on screen, before you download anything. You then get an Excel workbook, or a Xero bill-import CSV of the same rows.

Built for scans

Send whatever your scanner gives you.

We tested this against a known-correct control rather than relying on how it looked.

What we did

Twelve documents were printed on paper, scanned back in at 300 dpi and again at 150 dpi, and read three ways: from the original PDF, from the high-quality scan, and from the low-quality one.

What came back

Every date, reference, net, VAT, gross, currency and document type matched across all three readings, with no errors in any figure. The 484 KB scan read as accurately as the 8.1 MB one, so there's no need to worry about resolution.

bundle-E-scanned.pdf 16 pages boundaries found 12 / 12 documents invented 0 documents missed 0 native PDF vs 300 dpi all fields match native PDF vs 150 dpi all fields match 300 dpi vs 150 dpi all fields match file size 8.1 MB → 484 KB accuracy unchanged
FAQ

Converting a merged bundle

Do I have to split the PDF into individual invoices first?

No, that's the job. Give it the bundle as it came off the scanner and it works out where each document starts and ends, including where there's no text layer to help it.

What if a single invoice runs over several pages?

It stays as one invoice and one row. A description that wraps onto a second line is still a single item, and a figure mentioned in the body text isn't treated as a charge.

How long does a 300-page bundle take?

We've measured 3.5 seconds per document on the text path and 4.3 seconds when it has to be read as a scan — 15 documents end to end in 52 seconds. We have not yet timed a bundle at 300 pages, so we're not going to quote you a figure for one. Send us yours and we will time it properly.

What happens to a receipt or a council tax bill in the middle of the pile?

It is read, typed, and reported like everything else — and left off the bill. Only invoices and credit notes are charged for.

What if an invoice in the bundle doesn't add up?

Every figure printed on the page is written out as printed, the failed sum is described in words on the row, and the row is repeated on its own sheet. A supplier's arithmetic is never quietly corrected, and an empty field means the figure wasn't printed on the invoice at all.

What comes out at the end?

An Excel workbook — sorted oldest first, autofilter on every column, unverified rows on their own sheet — and optionally a Xero bill-import CSV of the same rows, with the nominal code column left blank for you to fill down.

Send us a bundle

For UK accountants and bookkeepers, by an RICS Registered Valuer who signs off numbers for a living.

Request access