Compare Two PDFs and Find Every Changed Word

Someone sends back "the final version" of a contract. It looks the same as the last one. Page count is identical, the logo is in the same place, and the email says "only minor edits". Do you read all 24 pages again to find them?

A word-level comparison does that reading for you. It lines up the text of both versions, marks what was added and what was removed, and leaves everything that matches alone. This post shows how to use the PDF comparison tool, what it is genuinely good for, and where its blind spots are, because a diff you trust too much is worse than no diff.

What a PDF comparison actually does

Compare PDFs does two things. First it pulls the text out of each file, page by page, using the same engine that renders PDFs in browsers. Then it runs a word-level diff on the joined text. The algorithm finds the longest sequence of words that both versions share, treats everything else as an edit, and outputs a single stream where:

• unchanged words appear normally,
• words only in the second file are highlighted green,
• words only in the first file are highlighted red with a strikethrough.

The result reads like tracked changes in a word processor, though it is generated from the final documents, not from edit history.

How to compare two PDFs

1. Open Compare PDFs.
2. Drop the original file into the first box. The label reads "Original PDF" until you choose a file.
3. Drop the changed file into the second box.
4. Click Compare. Processing happens in the tab, and the result appears below.
5. Read through the highlighted text. Look at red first (what was taken out), then green (what was put in).

If the two files are very large, you can narrow the job. Use Split PDF to pull out the same section from each version, then compare just those. That is faster and the output is easier to read.

Comparing text that is not in a PDF

Sometimes the content is in an email, a spreadsheet cell or a code block. In that case use Text Diff. Paste the original into the left box and the changed text into the right, and the comparison updates as you type. It uses exactly the same word-level function as the PDF tool.

A handy trick for Word documents or web pages: select all, copy, paste into Text Diff. For a PDF where you want the raw text for another purpose, Extract Text gives you the per-page text to copy.

What it is good at

• Spotting numbers that moved. A changed price, date or percentage is a single red word next to a single green one.
• Finding inserted or deleted clauses. A whole paragraph appears as a block of green or red.
• Checking a re-export. After regenerating a PDF from source, you can confirm that only the intended sections changed.
• Proofreading translations or edits. You see what a colleague changed without asking them for a list.

What it cannot see

This is the part most pages do not tell you, so read it before relying on a comparison.

• No images. A replaced logo, signature, chart or photograph is invisible to a text diff.
• No formatting or layout. Bold, font changes, colours, margins and column order are ignored unless the words themselves moved.
• No scanned pages. Without a text layer, there is nothing to compare. A scanned contract compared with a text-based one will show everything as changed or nothing at all.
• Reading order quirks. PDFs store text in drawing order, which is not always the visual order. Multi-column pages and tables can be extracted in a surprising sequence, producing noisy differences even when nothing changed.
• Hyphenation and line breaks. A word split across two lines may be read as two words.
• Size. The diff builds a table whose size is roughly the number of words in one file times the number in the other. That is fine for dozens of pages and not designed for a whole book.

If the stakes are high, such as a legal agreement, use the comparison to find changes quickly but still read the sections that matter. A tool that shows no difference means no difference in the text it could read, not proof the files are identical.

Where current tools fall short

Dedicated comparison tools tend to be paid, require uploading both documents to a server, or cap pages and file counts on free plans. Comparing two versions of a contract is exactly the case where you would rather the files stay private, and a server round trip adds a stranger to a confidential conversation. For the reasoning, see is it safe to upload confidential PDFs.

The browser version trades away the heavy machinery. There is no side-by-side page rendering, no image comparison, no export of a redline PDF. What you get is a clear text diff that costs nothing and leaves nothing behind.

Everyday use cases

• Contract revisions. Confirm the "minor edits" really are minor.
• Terms and policies. Compare last year's terms of service with the new version you were asked to accept.
• Academic drafts. Check what a co-author changed between two exported drafts.
• Quotes and proposals. Catch a changed quantity or line item before you sign.
• Manuals and specs. See what moved between revision A and revision B of a technical document.

A worked reading of a comparison result

Suppose version one of a quote says "Delivery within 10 working days, total 4,500" and version two says "Delivery within 14 working days, total 4,950, payment due on receipt". The output reads as one stream: the words "Delivery within" are plain; the word 10 is red with a line through it, and 14 follows in green; 4,500 is red and 4,950 is green; then ", payment due on receipt" appears all in green because it is new.

Three habits make this easy to read. First, scan for red-green pairs, which are replacements, such as a changed number. Second, scan for long green runs, which are insertions of new clauses. Third, scan for long red runs, which are deletions that someone may hope you will not notice.

Be careful with small tokens. A change from "may" to "shall" is a single word and easy to miss in a long page, but the highlighting makes it stand out. Equally, an unchanged sentence moved to a different page appears as a red block in one place and a green block in another, so a lot of red and green does not always mean new content.

If the output is mostly noise, check that you loaded the right files, and that both are text-based rather than scans. Then try comparing only the sections that matter by extracting them first.

Tips for reliable comparisons

• Keep order consistent. Original first, changed second, every time.
• Compare like with like. Two exports from the same tool give cleaner results than a PDF versus a scan.
• Trim headers and footers if needed. Changing page numbers or dates in a footer appear as edits on every page. Splitting the pages you care about helps.
• Check unusual layouts by eye. Tables and columns may extract in a strange order.
• Do not paste secrets you do not need. Even though nothing is uploaded, keep working copies tidy.
• Re-run after fixing. If you recreate a PDF to remove an unwanted change, compare again to confirm.

Related reading

If your files need trimming first, read how to split a PDF into pages. For the opposite job of joining files, see merge PDFs without uploading. For text-only changes, Text Diff is the quickest route, and more PDF utilities are collected under PDF tools.