Finding what changed between two versions
A contract comes back from the other side marked "minor amendments". A policy is reissued. A quote is revised. Reading both documents side by side to find the difference is slow and unreliable — the changes that matter are usually small, and a single word can reverse a clause. This compares the text of two PDFs line by line and shows exactly which lines were added and which were removed.
Both files are read in your browser. Nothing is uploaded, which matters when the two documents are drafts of an agreement that has not been signed yet.
How the comparison works
Text is extracted from each document and grouped into lines by vertical position, then matched with a longest-common-subsequence diff — the same approach source control uses. Lines present in both are treated as unchanged however far apart they have moved; the rest are reported as added or removed.
| Marker | Meaning |
|---|---|
| − red | In the original, not in the revision |
| + green | In the revision, not in the original |
| plain | Identical in both — hidden unless you ask to see the whole document |
An edited line shows as one removal followed by one addition, since a changed line is not the same line. Reading the pair together tells you what was rewritten.
What it compares, and what it does not
- It compares words. That is what people almost always mean by comparing two contracts, and it is the part that carries legal weight.
- It does not compare appearance. A changed font, a moved logo, a different margin or a new page break produce no difference at all, because the words are identical.
- It does not compare images. A replaced photograph or a redrawn chart is invisible to a text diff.
- It does not read scans. A PDF with no text layer compares as empty; the tool says so rather than showing a blank result. Run it through OCR PDF first, and bear in mind that recognition errors will then appear as differences.
- Tables come out as lines of text, because a PDF has no cells. Changes are visible but the column they belong to is not.
Reading the result sensibly
Leave the spacing option on unless you have a reason not to. A PDF stores the position of each run of glyphs rather than a line of text, so re-flowed paragraphs, justified spacing and soft line breaks differ constantly between two exports of the same content. Ignoring spacing removes a large volume of differences that are not changes.
Even so, expect some noise. Headers, footers and page numbers appear inline and shift when pagination changes, so a document that gained a paragraph on page 2 may show every later page header as a difference. Read the run of changes, not the count.
When the two documents are very long
The diff compares every line of one document against every line of the other, so memory grows with the product of the two lengths. The tool caps a comparison at four thousand lines each, which covers most contracts and reports. Beyond that, split both documents at the same point with Split PDF and compare the parts — the result is identical and each part finishes quickly.