You have two PDFs that should be similar — last week's contract draft and this week's revision, the report you sent and the one you got back, the proposal version you signed off on and the one the client says was final. Somewhere in those pages a clause changed, a number moved, a paragraph went missing. This guide covers how PDF comparison actually works under the hood, the three diff modes that matter, and how to find changes in a long document without scrolling through every page.
Why PDF comparison is harder than a plain text diff
A text file diff is straightforward: line A on the left, line A on the right, character-by-character compare. PDFs are not that simple. A PDF is a collection of positioned visual elements — text glyphs at specific X/Y coordinates, embedded images, vector shapes — laid out on a page. The same document re-saved by different software can shift a paragraph by a single pixel without changing its content. A version that swapped its embedded font will look identical but compare differently at the byte level.
So a useful PDF comparison tool can't just run a binary diff. It has to ask the harder question: which pages look different to a human reader, and how? That requires three things in combination:
- A pixel-level visual comparison — render each page to an image and compare pixel-by-pixel to surface visual changes the eye would notice.
- A text-level comparison — extract the actual text from each page and compare it as text, so a font change that doesn't alter the words doesn't register as a difference.
- Per-page status tracking — page-by-page summary so you can jump straight to the changed pages without scrolling through identical ones.
A good compare tool gives you all three on the same screen.
The three compare modes
Our Compare PDFs tool lets you switch between three views of the same comparison, each best for a different kind of question.
Visual Diff (3 panels)
Renders PDF 1, PDF 2, and a pixel-overlay panel side by side. The overlay panel renders one PDF beneath the other and tints the regions where they differ — added pixels in green, removed pixels in red. The overlay is the fastest way to spot a moved paragraph, a deleted image, or a shifted table.
This is the right view when you care about what the change looks like on the page — useful for design review, layout proofing, and "did this signature get added in the right place."
Side-by-Side (2 panels)
Two-panel view — PDF 1 on the left, PDF 2 on the right, scrolled together. No pixel overlay, just both pages rendered cleanly so you can eyeball changes the way you would on a physical desk.
Best for contract review where you want to read both versions in parallel and spot wording changes that would be hard to read in an overlay.
Text Diff (per-word)
Extracts the text from each page and renders a word-by-word comparison: added words highlighted in green, removed words in red, unchanged words in their original colour. This is the right view when the document is mostly text and the question is "what wording actually changed."
The Text Diff view needs a text layer in the source PDF. Scanned (image-only) PDFs have no text to extract, so the Text Diff tab won't help there — run those through our OCR tool first, or stick to the Visual / Side-by-Side modes for scans.
The badge system: how to find changes without scrolling
When two long PDFs land in the tool, you don't want to scroll through 200 pages looking for the one that changed. So the moment both files are loaded, the tool runs a pixel-diff scan in the background — comparing each page of PDF 1 against the same-index page of PDF 2 and producing a percentage-difference score.
Each page gets a colour-coded badge in the left sidebar:
- Green — pages are identical (or visually indistinguishable, e.g. only a single-pixel anti-aliasing difference).
- Orange (with %) — pages are different, with the actual percentage shown so you can prioritise the most-changed pages first.
- Red — page exists in PDF 1 but not in PDF 2 (PDF 2 dropped this page).
- Blue — page exists in PDF 2 but not in PDF 1 (PDF 2 added this page).
Click any badge in the sidebar to jump straight to that page. Or click Next Diff in the toolbar to skip every identical page and jump straight to the next changed one — typically the single most useful button for reviewing mostly-unchanged revisions.
Lazy rendering: why the tool feels fast on a 500-page PDF
If the tool rendered every page upfront at full fidelity, opening a 500-page comparison would freeze the browser for a minute. So it runs in two passes:
- Thumbnail-sized pixel-diff scan of every page upfront (about 30 ms per page on a typical laptop, so even a 500-page document scans in under 30 seconds). This is what drives the sidebar badges.
- High-fidelity per-page rendering of only the page you're currently looking at — re-rendered the moment you click Next, Prev, the sidebar, or Next Diff.
The first 20 pages render eagerly so you can start reviewing within a second of upload. Pages 21+ render on-demand as you navigate. A "render token" pattern cancels stale renders if you click through pages fast, so rapid navigation doesn't produce flicker or out-of-order content.
The net effect: a 500-page comparison opens just as quickly as a 5-page one, and feels responsive even on a phone.
Limitations to know about
- Page-to-page alignment is by index. If PDF 2 inserted a new page at position 5, the original pages 5, 6, 7, 8 in PDF 1 will be compared against the inserted page and the shifted originals in PDF 2 — every subsequent page will register as "different" even though the content shifted rather than changed. This is a known limitation shared with most PDF diff tools; the alternative is structural alignment, which is expensive and produces unreliable matches on real-world inputs. The percentage column helps spot this case: shifted content typically registers as 80–95% different rather than the 5–30% you'd see for a real edit.
- Visual Diff is per-pixel. A different anti-aliasing setting, a different export-to-PDF software, or even a re-save in Adobe Acrobat can produce sub-pixel rendering shifts that the visual diff registers as changes. Use Text Diff alongside to confirm whether the text actually changed.
- Text Diff requires a text layer. Scanned PDFs need OCR first (see above).
Privacy: nothing leaves your browser
Both PDFs are loaded entirely into your browser memory, rendered in-browser, and diffed locally. Nothing is uploaded to our server. No signup. No watermark on anything. Close the tab and both files disappear from memory — there's nothing on our side to clean up because nothing was ever on our side.
This matters for the typical compare workflow: confidential contracts, draft proposals, internal financial reports. The same legal review where you'd hesitate to drop a PDF into a generic web converter is exactly the workflow Compare PDFs is built for.
Walk-through: a real comparison
- Open PDFGrover Compare PDFs.
- Drag the original PDF into slot 1 (left) and the modified PDF into slot 2 (right).
- Wait for the upfront scan to populate the sidebar badges — typically a few seconds.
- Click Next Diff in the toolbar to jump straight to the first changed page.
- Toggle between Visual / Side-by-Side / Text Diff to see the change from the angle that makes it clearest.
- Repeat Next Diff to step through every change in order.
That's the whole flow. No upload, no account, no watermark.
Further reading
- Tool page: Compare PDFs — interface, the three compare modes, sidebar navigation
- Related: Merge PDF — once you've identified which version is right, combine the good pages
- Related: Extract Text from PDF — pull plain text if you only need the words for a side-by-side review
- Related: OCR PDF — make a scanned PDF text-searchable before running Text Diff
Compare two PDFs now — free, no upload, runs entirely in your browser.