Split a PDF by Keyword: One File per Invoice, No Upload

Your accounting software just exported a single 200-page PDF: the quarter’s invoices, one per page. You now need to send each invoice to a different client — in other words, turn that block into 200 separate files. That’s exactly what text-based splitting is for: rather than cutting “every N pages” or “at bookmarks”, you cut at every page containing a marker word — here, “Invoice #”.

How text-based splitting works

The mechanism is easy to grasp — and two of its details matter a lot in practice:

1. The marker page opens the next segment. If the word “Invoice” appears on pages 1, 40 and 87, you get three files: pages 1-39, 40-86 and 87-200. Each invoice therefore starts with its title page — the intended behavior.
2. The search is case-sensitive. The engine tests for the exact string: “INVOICE”, “Invoice” and “invoice” are three different markers. Pick the most stable string possible (“Invoice #”, “CERTIFICATE”, “Separator page”) rather than an ambiguous word.
3. Page 1 never triggers a cut — a marker on the first page is simply ignored, since there is nothing before it.
4. No occurrences? The document is returned as-is, with no error.

Brother PT-H110 — portable label makerDurable laminated labels: filing that lasts.affiliate

The right hardware

Brother PT-H110 — portable label makerDurable laminated labels: filing that lasts.affiliate Oxford Top File+ — pack of 10 foldersThree flaps and elastic: the standard format to file a case.affiliate

Step-by-step workflow

Back to our invoice batch:

1. Check the marker. Open the PDF and find the text that appears at the top of each invoice and only there — a word that also shows up in the terms and conditions or the footer will create spurious cuts.
2. Run the text-based split with that exact string. Processing happens in the browser: the file is never sent to a server (50 MB and 1,000 pages maximum).
3. Download the ZIP containing one PDF per segment, then rename the files if needed — output names follow the segment order, not the marker content.

Tip: if your batch has no marker word but the invoices have a fixed length (one page each), splitting every X pages does the same job with a single parameter. And if the source document has per-chapter bookmarks, bookmark splitting or outline-based extraction is even cleaner.

Limits to know before you start

• A pure scan won’t work. Detection reads the PDF’s text layer. A scanned document is an image: nothing to search. Our local OCR produces raw text — it does not inject a text layer back into the PDF, so it won’t prepare a scan for this tool. Text splitting targets digitally generated PDFs (software exports, print-to-PDF output) or files already OCR’d elsewhere.
• The marker must be unique to starting pages. Test it with your PDF reader’s search: it should only appear on pages that open a new document.
• Case and spacing matter. “Invoice #” with two spaces won’t match “Invoice #”.
• Output is a ZIP, one file per segment. If you actually wanted a single PDF containing certain pages, page extraction is what you need.

In short

For a homogeneous batch — invoices, certificates, payslips, statements exported as a single PDF — text-based splitting turns an indivisible block into files you can name and send, with no upload and no page-by-page handling. The key is choosing the marker: precise, stable, unique to starting pages. Ten seconds of checking before the split prevents a ZIP full of incoherent segments.