A vendor sends back a “revised” contract. A colleague emails a “final” version of a report you already read once. Somewhere in there, something changed — a clause got reworded, a number got adjusted, a sentence quietly disappeared — and you need to know exactly what, without re-reading the whole document line by line hoping you notice.
Why eyeballing it doesn’t work
Side-by-side skimming catches big changes and misses small ones. A single word swapped inside a long paragraph — “will” changed to “may,” a date bumped by a day, a clause deleted from the middle of a list — reads as identical at a glance, especially in a document with dense formatting or legal boilerplate. The changes that matter most in a contract or a policy document are usually the ones written to blend in.
The fast way
- Open Compare PDF and upload the original file and the revised one.
- It compares them page by page automatically.
- For any page that differs, expand it to see the diff.
That’s the whole workflow — no accounts, no waiting on a shared review tool, and nothing gets uploaded anywhere, since the comparison runs entirely in your browser.
What the red and green actually mean
For each page that differs, you get the original page with the removed text boxed in red and the revised page with the added text boxed in green, plus a word-by-word text diff underneath showing exactly which words were struck out and which were inserted. Because the comparison works at the word level, it holds up even when a paragraph reflows onto a different line after an edit — it’s matching wording, not position on the page.
Pages that are identical are marked as matching and collapsed by default, so you’re not stuck scrolling past forty unchanged pages to find the two that actually moved. If the two files don’t have the same number of pages, extra pages in either one are flagged as “only in this file” instead of being awkwardly compared against nothing.
When the word diff isn’t enough
Text-based highlighting is precise for wording, but it won’t catch everything. A swapped image, a shifted table, or a color change isn’t a change in words, so it won’t show up as red or green text. For that, each page also has an optional “highlight what changed on the page” view, which overlays both versions and flags every pixel that differs — useful for catching a replaced diagram or a moved logo that a word diff would miss entirely.
Two things worth knowing about that overlay before you rely on it. First, it needs both pages to render at the same pixel dimensions — if the revised file uses a different paper size, you’ll get a note that they can’t be overlaid rather than a diff. Second, it’s built for spotting a few small, localized changes, not for pages where the wording reflowed. If enough text shifted that most of the page moved, the overlay would just highlight nearly the whole page as “different,” which isn’t useful — so instead of showing that noise, it won’t display at all.
And if a page is a scan or otherwise has no extractable text at all, the tool falls back to that same pixel overlay automatically, since there’s no wording to compare in the first place.
What it won’t catch
A few things sit outside what a page-by-page PDF diff — this one included — can tell you.
Pages that were inserted, deleted, or reordered partway through. The comparison lines up by position: page 1 against page 1, page 2 against page 2, and so on. If someone adds a page in the middle of the revised document, everything after that point lines up one page off from where it should, so it all gets flagged as “different” even though most of it didn’t actually change — it’s just being compared against the wrong page now. If you know a page was inserted somewhere, factor that shift in before you trust every “different” flag after it.
Comments, highlights, and form field values. The diff works from the text and pixels actually printed on the page, not the annotation layer sitting on top of it. A sticky-note comment or a filled-in form field added between drafts won’t turn up as a change.
Anything that isn’t a PDF. This compares PDF to PDF only. If one side of the comparison is a Word file, export or print it to PDF first — there’s no built-in PDF-to-Word comparison.
None of this makes the tool unreliable — it just means “different” always means the wording or pixels changed, never that a page moved or a comment got added. Knowing which of those three explains a flagged page is usually obvious once you look at it.
Where this actually gets used
- Contract redlines — confirming a “minor cleanup” revision from a counterparty really is minor, before signing.
- Policy or SOP updates — checking exactly what changed between the version everyone already trained on and the new one.
- Manuscript or report revisions — seeing what an editor or co-author actually touched, beyond what track-changes comments called out.
- Government or vendor filings — catching a quietly modified figure or deadline in a resubmitted document.
If you only need to confirm a document hasn’t been tampered with rather than review a specific revision, Compare PDF still works well as a first pass — run it, and anything with no differences flagged is confirmed identical, word for word.