Comparing Two PDF Versions: Finding What Changed Before You Sign

Someone sends back a contract with "just a few small changes". The few small changes are not always the only ones. A figure can move, a clause can be reworded, a date can slip, and the version you sign may differ from the one you read. Comparing the two files is the quick way to find out, and it is worth doing before anything is signed or sent on.

What the word comparison does

The compare tool has two modes. Words mode takes two PDFs and reads the text layer of each: the words the file stores for its pages, in reading order across all of them. It lists the words that were removed from the original and the words that were added in the revised version, and says which page each change is on. A change to one figure looks like one removed word and one added word, side by side, which is easy to check at a glance.

Because the comparison works on the document as one stream, an inserted page does not make every page after it look changed. Only the words that are actually new or gone are listed. You can download the same list as a text file, which is handy for a record of what was agreed.

What it treats as a change

Words are compared exactly as the file stores them. Capital letters, punctuation and the order of the words all count, so "Net 30" and "net 30" differ. Spaces and line breaks do not count, so a paragraph that is only re-wrapped to a different line width shows no change.

Two patterns can look like changes when nothing of substance has changed. A word split by a hyphen at the end of a line can be stored as two pieces in one version and one piece in the other. And a paragraph that was moved shows as removed in one place and added in another, which is accurate about where the words went but can look like two edits. Read the report with that in mind.

What the word comparison cannot see

Words mode is about words. It does not see a changed image, a signature that was replaced, a table redrawn with the same numbers in a new layout, a highlight someone added, or a page that was reordered without any words changing. A reviewer who only reads the word report could miss those, which is what Pictures mode is for.

Scanned pages are a bigger gap for Words mode. A scanned page is a picture of text, with no text layer, so there are no words to compare. The report names those pages as not compared, so you know to check them another way. The guide to scanned and text PDFs explains how to tell which kind you have.

Comparing the pages as pictures

Pictures mode draws every page of both files in your browser and compares them as images, page 1 with page 1, page 2 with page 2. It lists the pages that look different, with the share of each page that changed, and you can open one to see what is only in the original (in red) and only in the revised version (in green), with amber where only a colour changed. That finds a replaced signature, a swapped image or a moved table, and it works on scanned pages too.

It has limits of its own. Pages are compared at about 108 DPI, so a change of a pixel or two can be missed, and "looks the same" means the pictures matched at that size, not that the files are identical. Pages are paired by number, so a page added near the front makes every later page look different. Pages of different sizes are listed but not compared. And a file saved again by a different program can draw its text slightly differently even when nothing visible changed. It shows where pages differ; it cannot read what a scanned change says, so look at the marked areas yourself.

A sensible way to use it

1. Keep the version you last agreed to, and the version you were sent. Give them clear names so the report says which is which.
2. Run Words mode and read the summary first. A count of zero means no text changed; it does not mean nothing changed.
3. Go through the list page by page. Pay most attention to numbers, dates, names and the words "not", "shall" and "may".
4. Run Pictures mode too, and open each page it lists. If the word report named pages with no text, this is how you check them.
5. Download the report and keep it with the file you signed or sent, so the record of what changed travels with it.

The comparison runs in your browser. Both files are read on your device, and the report is made there, so the documents are never sent to a server.