The Complete PDF Toolkit Guide: Merge, Convert to Word, Export Images, Compress—All in the Browser
PDF Is the Universal Format—But Editing Is Another Story
Contracts, papers, invoices, product manuals—almost any formal document eventually becomes a PDF. Its greatest strength is exactly "finalized = fixed": open it on any device and the layout never shifts.
But that strength is also the trouble: editing two lines in Word is trivial, yet merging pages, extracting an image, or converting back to an editable format in PDF stops a lot of people. In the years we've built the PDF toolkit, we found demand concentrates on a handful of tasks. This article explains them all at once, so you can find the steadiest path between "can't edit" and "getting burned by an online tool."
First, Understand: Why PDF Is So Hard to Edit
To get why "PDF to Word" is so hard, you need to know PDF isn't a word-processor format. Internally a PDF is organized like this:
- Objects: text, fonts, images, paths—each an independent object.
- Content Stream: each page is a string of drawing commands saying "draw this glyph at (x,y), in this font, this size."
- Page Tree: links the pages together.
The key point: a PDF stores "what it looks like drawn," not "this passage is a paragraph, that sentence is a heading." It doesn't inherently know which are paragraphs, which are tables, which are heading levels. So converting back to Word means software must reverse-guess the semantic structure—that's why complex layouts break. Understand this and you'll stop raging at "imperfect Word conversion."
Merge: Combine Several Into One
The #1 most common need: a client sends three files, the scanner spits out twenty pages, and you still need to attach your signed page at the end. You want one PDF to send back.
PDF Merge is straightforward—drag the files in, reorder by dragging, click to concatenate in order. Runs entirely locally in the browser, no upload.
Good fits:
- Contract body + signature page + appendix into one reply
- Scan batches generated separately, combined into one volume
- Multiple departments' reports unified into one deliverable
- Paper body + references + appendix consolidated
Watch out when merging:
- Order: confirm order by dragging before merging; sending out scrambled page numbers is the most common low-level mistake.
- Mixed page sizes: A4 and Letter mixed leaves white margins on some pages; unify first.
- Bookmarks/links: simple concatenation usually doesn't preserve the original bookmarks and internal hyperlinks; use desktop Acrobat if that matters.
Convert to Word: Most Asked, Most Disappointing
"Help me turn this PDF into Word" is a frequent request, but also the hardest to manage expectations on.
PDF to Word uses cloud document-intelligence recognition (Alibaba Cloud Document Intelligence) to restore layout into DOCX at high fidelity. Here's the trade-off that must be stated: calling cloud recognition means the file must be uploaded to the cloud. So it fits final drafts, public, or non-confidential content—not contracts with sensitive info.
More important is expectation. Real performance by document type:
| Document type | Fidelity | Note |
|---|---|---|
| Single column, clean font, standard layout | Very high | Almost no fixing needed |
| Standard tables | High | Lines and content mostly align |
| Multi-column (papers/magazines) | Medium | Column order occasionally wrong, manual fix |
| Cross-page large tables | Medium-low | May break rows, misalign |
| Math formulas, chemical equations | Low | Often become images or garbage, re-key |
| Handwritten annotations | Very low | Unrecognizable, manual |
Treat it as "saves 80% of the re-layout work," not "one-click perfect." Verify key pages, then do the remaining edits—saves a lot of rework.
Table-heavy docs can use PDF to Excel, which does table-layout recognition and exports XLSX—easier than picking tables out of Word.
Export Images: Page by Page
Sometimes you don't want to edit, you want to "use" a page: send a paper page to a group, make a product page a thumbnail, insert a scan page as an image into PPT, upload an invoice page to the reimbursement system.
PDF to Images and PDF to JPG export per page as PNG/JPG/WebP, auto-zipped, with an adjustable clarity multiplier. Retina/print needs a higher multiplier to stay sharp—the multiplier is essentially render DPI scaling: 1× ≈ 72–96 DPI, 2× ≈ 144–192 DPI; for print use 2× or more.
How to choose:
| Need | Choose |
|---|---|
| Crisp, editable later, has text | PNG |
| Just to view/share, small size | JPG |
| Balance size and clarity, modern env | WebP |
These operations also run locally, no upload.
Compress: Especially Effective on Scans
Email caps attachments at 20MB, WeChat chokes on big files, cloud upload is slow—that's when you need PDF Compress.
Its principle is to rasterize pages and re-compress them as JPG into an image-type PDF, often cutting size by more than half. But be clear: this works wonders on "scanned/image-type PDFs" because they're already a pile of images; on "text-type PDFs" (selectable text, small size) there's almost nothing to squeeze, so it barely shrinks.
So the rule is simple:
Is the PDF large?
├── Yes, looks like a photo (scanned/image-type) → compress, big effect
└── No, text is selectable (text-type) → don't bother, no use and slower
Advanced option: Linearize (Web optimized). If your PDF previews online, linearization lets the browser show the first page while still downloading—much better UX.
A Whole Set of Peripheral Features
| Feature | Tool | Common use |
|---|---|---|
| Split | /en/pdf/split | Pull a few pages from a large doc |
| Rotate / delete pages / crop margins | /en/pdf/rotate, /en/pdf/delete-pages, /en/pdf/crop | Fix scan skew, remove black edges |
| Encrypt / decrypt | /en/pdf/encrypt, /en/pdf/decrypt | Set open password on confidential files |
| Text / stamp watermark | /en/pdf/watermark, /en/pdf/stamp | Internal proofs, copyright marks |
| Page numbers / header-footer | /en/pdf/page-numbers, /en/pdf/header-footer | Add standard page numbers to deliverables |
| Extract text | /en/pdf/extract-text | Export text-type PDF to txt |
| Reverse order / metadata | /en/pdf/reverse, /en/pdf/metadata | Special organizing needs |
All these features run locally in the browser; files never leave your machine.
The Privacy Trade-off, Stated Clearly
This is the most overlooked yet critical point:
- Local, no upload: merge, image export, compress, split, rotate, encrypt, watermark, text extraction—all run in the browser, no upload.
- Needs cloud upload: Word and Excel conversion go through cloud document intelligence; files must be uploaded to recognize.
So the principle: confidential contracts, documents with ID numbers or bank cards—don't use cloud conversion. Either desensitize locally first (use delete pages to drop sensitive pages before processing), or stay within local-only features.
Common myth: "well-known online PDF sites are safe." Most do claim not to retain, but the file still passes through their servers. For sensitive files, local tools are the steadier choice.
Two Common Workflows (Copy-Paste Ready)
Scenario A: client sends scattered pages, wants back one tidy PDF
Merge /en/pdf/merge → export images to preview /en/pdf/pdf-to-images → compress if too big /en/pdf/compress → send.
Scenario B: scanned contract needs editable version
Desensitize locally first (delete sensitive pages) → upload to Word /en/pdf/to-word → fix a few spots in Word → encrypt locally /en/pdf/encrypt → send back.
Scenario C: reimbursement needs a single invoice image
/en/pdf/pdf-to-jpg export per page at 2× → pick the right page → upload to reimbursement system.
Command-Line Options (for Developers)
When you don't want a web tool and have large batches, open-source CLI tools are handy:
# qpdf: merge (local only, no re-compression)
qpdf --empty --pages a.pdf b.pdf c.pdf -- out.pdf
# qpdf: compress / linearize
qpdf --optimize --object-streams=generate in.pdf out.pdf
# pdftk: extract a few pages
pdftk in.pdf cat 1-3 7 end output part.pdf
# ghostscript: compress scanned PDF to small
gs -sDEVICE=pdfwrite -dCompatibilityLevel=1.4 \
-dPDFSETTINGS=/ebook -dNOPAUSE -dBATCH \
-sOutputFile=out.pdf in.pdf
Note: CLI is also local, great for batch automation; but cloud recognition (the high-fidelity Word restore) generally isn't available in CLI tools—that's the strength of document intelligence.
Common Mistakes
| Mistake | Consequence | Fix |
|---|---|---|
| Expect perfect Word conversion | Complex layout breaks, wasted wait | Treat as 80% automation, budget manual fix |
| Force-compress text-type PDF | Barely shrinks, slower | Only scans compress well |
| Forget to encrypt confidential | Can't recall after sending | Local encrypt /en/pdf/encrypt |
| Cloud-convert confidential | Leak risk via server | Desensitize first or local-only |
| No order check before merge | Scrambled page numbers | Confirm order by dragging |
| Too-low image multiplier | Blurry on print/retina | Use 2× or more |
FAQ
Will formulas and tables break after Word conversion? Standard layouts are stable; multi-column, cross-page tables, and formulas may need manual fixes. Verify key pages before big edits.
Does compression lose content? No text info is lost (image-type compression re-rasterizes pages), but image quality drops slightly. Scans are usually visually identical after compression.
Forgot the encryption password? Local encryption uses the password you set; the tool doesn't store or upload it. If forgotten, only the password itself can recover it—no backdoor. Store it in a password manager (see the password security guide).
PNG or JPG for image export? PNG for crisp/editable-later; JPG for view/sharing with small size; WebP balances both but older software may not read it.
Free online tools vs local tools? The core difference is "does the file pass through a server." Local tools run entirely in your browser—steadier privacy; online tools are convenient but the file reaches their server.
One-Line Decision
Need to process a PDF?
├── Several into one → /pdf/merge
├── Edit text → /pdf/to-word (note: cloud upload)
├── Just tables → /pdf/to-excel
├── Images / thumbnails → /pdf/pdf-to-images
├── File too big → /pdf/compress (scans most effective)
├── Just a few pages → /pdf/split
└── Confidential → local encrypt /pdf/encrypt, skip cloud
Keep this guide. Next time someone throws you a "help me with this PDF," pick the tool by the branch and finish in minutes. To compress or convert the exported images further, see the image compression guide.
Try these browser-local tools — no sign-up required →