PDF power tools
Compare two PDFs
Find the words that changed between two versions of a document. Your files are processed locally and are never uploaded to Novus PDF Studio.
Find the words that changed between two versions of a document.
Document 1
Earlier version
The original, or whichever copy came first.
Document 2
Later version
The revised copy. Changes are reported against document 1.
The order decides the answer: a word only in document 2 is an addition.
Loads 2 samples: quotation, revision 1 and quotation, revision 2, 2 pages in total.
Loaded in order: revision 1 is the earlier version. Three things changed between them.
Samples contain invented data only and are read from this site, not uploaded.
The two documents
Choose both documents to compare them.
Earlier version
Later version
The report compares the words on each page. It cannot see a moved logo, a changed signature image or a box drawn over a paragraph, and a scanned page has no text to compare. There is no highlighted side-by-side view.
Choose a PDF to begin.
About this tool
Somebody sends a contract back and says they made a small change. You have two files with similar names, a deadline, and no way to tell which words moved. Reading both is slow and unreliable, because the eye slides straight over a changed digit in a figure it has already read once.
This tool reads the words off each page of both documents and reports the ones that differ, page by page. It compares the earlier version you give it first against the later one you give it second, and the order matters: swapping them turns every addition into a removal. The result downloads as a plain text report, which is a thing you can paste into the email where the argument is actually happening.
It compares text, and only text. That is the half of comparison that catches a changed amount, a changed date or a reworded clause, which is what people open two versions of a document to find. It cannot see a moved logo, a different signature image or a box drawn over a paragraph. Both limits are printed at the bottom of every report rather than kept here.
How to compare step by step
- 1
Choose the earlier version
Drop the original into the first slot. The slots are labelled, because which document is which decides what the report calls an addition.
- 2
Choose the later version
Drop the revised file into the second slot. Use Swap if you filled them the other way round.
- 3
Decide about capitalisation
Leave case comparison on unless a heading has been restyled. Ignoring case is reported in the result, because folding it hides real changes as well as noisy ones.
- 4
Compare and read the report
The report names every changed page, lists the words removed and added, and ends with what it could not compare.
When to use it
- A contract comes back marked as a small change and you need to know which words moved.
- Two files have nearly the same name and you need to know whether they are actually different.
- A quotation has been revised and the figure matters more than the formatting.
- A policy document is reissued and you want the reworded clauses, not the whole thing again.
Tips
- Put the earlier version first. The report labels them Before and After, so a swap is obvious once you read it, and easier not to make.
- A report with no differences is a real answer, and it is worth keeping alongside the files.
- If one file is protected, unlock it first. A password-protected document cannot be read for comparison.
Good to know
- Text only. Images, colours, positions and formatting are not compared.
- There is no rendered page overlay; the result is a text report.
- Pages are aligned by position, not by content.
- A page with no text layer cannot be compared and is named as such.
- Reading order comes from the text layer, which is a guess on a multi-column page.
Compare questions, answered
Does it show the pages side by side with the changes highlighted?
No. A rendered overlay is a separate piece of work and it is not built. What you get is a text report: the changed pages, and the words removed and added on each. Saying this plainly matters more than it sounds, because a comparison that quietly compares less than you assumed is worse than no comparison at all.
What happens with a scanned document?
A scan is a picture of text, so there is nothing to compare. The report says the pages had no text layer rather than reporting them as unchanged. Those are opposite answers and confusing them is the failure that matters here. Recognise the text in both files first, and then compare the results, keeping in mind that recognition makes mistakes on exactly the characters you are checking.
Why does every page after page two say it changed?
Pages are compared by position: page three against page three. Inserting a page near the front shifts everything after it, and every shifted page then differs from the page it is compared with. The report detects that pattern and names it as a probable inserted or removed page rather than letting you read it as wholesale rewriting.
Can two identical-looking pages be reported as different?
Yes, in one situation worth knowing about. Reading order comes from each file's text layer, which is a guess on a page with columns or a complex layout. Two files that render identically can extract their words in different orders, and that reads as a change. If a report looks wrong in that specific way, the pages are worth looking at.
Are my documents uploaded?
No. Both files are read in this browser tab, by the same code that runs every other tool here. There is no upload endpoint. Nothing about either document, including its name, is recorded anywhere.
Privacy note: like every Novus PDF Studio tool, compareruns entirely in your browser. Files, passwords, and settings stay in this tab's memory and are discarded when you close it. There is no upload, no queue, and no server-side copy. Read more on the privacy page.