Accounting software, billing systems and bank portals love to export everything as one PDF. You asked for last month's invoices and got a single 212-page file with 87 invoices in it, some one page long and some four. Splitting by page numbers would mean paging through the whole thing and writing down where each invoice starts.
Split by text reads the text on every page and finds the boundaries for you. It can start a new file wherever a phrase appears, or wherever a value in a fixed spot (an invoice number, an account number) changes. Your file is read and split in the browser and isn't uploaded.
How a PDF stores the text this tool reads
A PDF made by software, as opposed to a scanner, contains real text: runs of characters, each placed at an exact position on the page in a particular font. That's what you select when you drag across a line in a PDF reader. When you open a file here, the tool reads those runs and their positions for every page, so it knows both what each page says and where on the page it says it.
Two quirks matter here. First, PDFs don't store words or lines, only positioned characters, so the spacing you see isn't always the spacing in the file. Some generators space out every letter, others drop spaces entirely. The tool compares text with whitespace ignored as a fallback, so "Invoice No" still matches when the file actually holds "InvoiceNo".
Second, a scan has no text at all. It's a picture of a page. If every page is like that, you'll see "This PDF has no text layer (it looks like a scan), so there is no text to split by." The "Make it searchable with OCR first." link takes the file straight to OCR PDF, which adds an invisible text layer (the guide to making a scanned PDF searchable explains how). Then come back and split.
The two modes, and which one you want
When text in an area changes is selected by default, and for batches of invoices or statements it's usually the right one. You drag a box around something that's different for each document and sits in the same place on every page, such as the invoice number in the top right corner. A new file starts whenever the text inside that box changes.
The nice part is what happens on continuation pages. If page 2 of an invoice has nothing in that spot, or repeats the same number, it stays with page 1. Empty boxes never start a new file. So a four-page invoice stays in one piece without you doing anything.
When a page contains text starts a new file on every page that contains a word or phrase anywhere on it. You type the phrase after "Split whenever this text appears:". Any pages before the first match become a file of their own.
This mode is simpler but blunter. Type Invoice into it and every page of a multi-page invoice that prints "Invoice" in its header starts a new file, and so does a cover letter that says "please find the invoice attached". It works best with a phrase that only appears on the first page of each document. If your statements print "Page 1 of 3" at the bottom, searching for Page 1 of often works well, since it only shows up on first pages.
"Case sensitive" applies to both modes and is off by default, so invoice matches INVOICE.
Splitting by an invoice number step by step
- Open Split by text and drop your PDF on the upload box. A password-protected file asks for its password first.
- Leave "When text in an area changes" selected. On the page preview, drag a box around the invoice number. Make it a bit larger than the text so a longer number still fits.
- Read the line under the preview, "Text in the box on this page:". It should show the number and nothing else. Use the arrows to check a few other pages.
- Check the summary, for example "Found on 140 page(s). 87 file(s) will be created.", and scroll the list of parts.
- Click the button in the header, "Split into 87 file(s)".
Several files download as a ZIP ending in -split.zip.
Getting the box right
Pick something that changes with every document. An invoice number is ideal. A customer name is risky: two invoices in a row for the same customer would end up in one file, because the text in the box didn't change between them.
If the box catches stray text on some pages (a date, or a "continued" label printed in the same spot), type the start of the real value into "Text should start with (optional)". With INV- in there, only text beginning with INV- can start a new file, and everything else is treated like an empty box.
The box sits at the same distance from the top-left corner on every page. That's fine when all pages are the same size. If the file mixes Letter and A4, or has a few pages turned sideways, tick "Scale box with page size" so the box grows or shrinks with each page.
If nothing is found at all, you'll be told "No text was found inside the box on any page. Draw it a little larger around the text." Usually the box was drawn too tight.
Naming the files after what's in them
In area mode, "Name files" is set to "by the text found", so each file is named after the text in its box: INV-20431.pdf rather than invoices-pages-14-17.pdf. Characters that aren't allowed in file names are cleaned out, and if the same text turns up twice the second file gets " (2)" added. Switch to "by page numbers" if you prefer the page-range names. Phrase mode always names files by their pages.
For more control, "Output file names" under "More options" takes a pattern with [FILENUMBER], [TEXT], [BASENAME], [CURRENTPAGE] and [TIMESTAMP]. Something like 2026-09_[TEXT] keeps a month's invoices sorted together. "Discard bookmarks" is there too if the PDF has any.
When to use something else
If the document has a bookmarks panel with one entry per chapter or section, Split by bookmarks is quicker and doesn't depend on the page text at all. If the documents are all the same length, say 50 two-page payslips, "Every N pages" in Split PDF is simpler. And if you already have separate files that only need better names, Rename PDF reads text from each file and renames it without splitting anything.
Also useful: OCR PDF · Split by bookmarks · Rename PDF
