Compare two PDFs
Put two versions side by side and find out what actually changed — which words were added, which were removed, and on which page. Neither file is uploaded.
Words, not pixels
There are two ways to compare documents and they answer different questions. A visual comparison overlays the pages and highlights anything that looks different, which is useful for checking a layout but nearly useless for checking wording: add one sentence on page 2 and everything after it shifts down, so a visual tool reports the rest of the document as changed.
A text comparison extracts the words and aligns them, so inserting a sentence is reported as inserting a sentence and the following pages come back clean. That is what this does, using a longest-common-subsequence alignment over words — the same approach code review tools use on source files.
What it will and will not tell you
| Change | Detected? |
|---|---|
| Words added, removed or reworded | Yes |
| Pages added or removed | Yes |
| Numbers or dates altered | Yes — they are words too |
| A replaced image or logo | No |
| Font, colour or spacing changes | No |
| Two scans of the same page | No — there is no text to read |
Why do it in the browser
The documents people compare are contracts, tenancy agreements, settlement drafts and policies — two versions of something where the question is precisely what the other side changed. Uploading both copies to a comparison service to answer that question is an odd trade. Here both files are read and aligned on your own machine.
Common questions
Does this compare the appearance or the words?
The words. That is the more useful of the two: a document reflowed by a single line will light up a pixel comparison from that point onwards while telling you nothing about whether any wording changed. Here an inserted clause is reported as an inserted clause.
Will it spot a changed image or a changed font?
No. Only text is compared. A replaced logo, a colour change or different formatting will not appear, and the result says so rather than implying it checked everything.
It says there is no text to compare.
Then at least one of the files is a scan — a photograph of a page contains no text, only pixels. Run both through OCR first and compare the results; the text layer is what makes comparison possible.
What about very long pages?
Word-by-word alignment is quadratic, so a page beyond a few thousand words is reported as changed without the word detail rather than freezing the tab. Most documents never reach that.
Are the files uploaded?
No. Both are read in your browser and compared there. For two versions of a contract or an agreement, that is usually the whole reason to avoid an online comparison tool.
Why are long unchanged stretches collapsed?
Because the point is to find what moved. Runs of identical text are summarised so the additions and deletions are actually visible, with a few words of context either side.