How to Compress a PDF Without Ruining It
Why PDF compression is a genuine trade-off, what it costs you, and when you should fix the source document instead.
Image compression is mostly free — you lose some detail nobody notices. PDF compression is not. Understanding why saves you from mangling a document that mattered.
What actually makes a PDF large
Text takes almost no space. A twenty-page text-only PDF is typically under 300 KB. If your file is 8 MB, the size is coming from one of three places:
- Scanned pages. Each page is a full photograph. A colour scan at 600 dpi is roughly four times the data of the same page at 300 dpi, and colour is three times greyscale — so a bad scan setting can produce a file twelve times larger than necessary.
- Embedded images at print resolution. A headshot or logo placed at 300 dpi when the document will only ever be read on screen.
- Un-subsetted fonts. The whole typeface embedded rather than just the characters used. Less common now, but it still happens with some exports.
The trade-off nobody explains
Every browser-based "compress PDF to 200 KB" tool, including ours, works the same way: it renders each page to an image, re-encodes that image at lower quality, and builds a new PDF from the results.
That hits the size target reliably. It also destroys the text layer. Your document becomes a stack of pictures. It is no longer selectable, searchable, or readable by any automated system.
For a scanned certificate, that costs you nothing — it was already an image. For a résumé being parsed by an applicant tracking system, it is close to fatal: the system reads the text layer, and you have just removed it.
Fix the source first
Before compressing, try these in order. Each preserves the text layer completely:
- Re-export at lower image quality. Word, Google Docs and most design tools offer a "web" or "smallest file size" export. This recompresses embedded images while leaving text as text — usually a 60–80% reduction on an image-heavy document.
- Remove or shrink the images. A résumé photograph is often the entire problem, and many employers prefer résumés without one.
- Rescan properly. 200–300 dpi in greyscale is ample for text documents. This does more than any compressor can afterwards, because you never create the excess data in the first place.
Only when those fail should you rasterise.
When rasterising is the right answer
- The document is already a scan, so there is no text layer to lose.
- A portal enforces a hard limit and rejects the file outright — a rejected application is worse than an unsearchable one.
- The document is going to a human who will only read it on screen.
Lossless operations
Not everything requires the trade-off. Merging, splitting, rotating, deleting pages and adding page numbers all edit the PDF structure directly. Text stays selectable, image quality is untouched, and file size barely changes:
A note on password-protected files
If your PDF is encrypted, no browser tool can process it, and we will not offer a way around that. Open it in your PDF reader with the password, save an unprotected copy, and work from that.
Size targets
If you have exhausted the alternatives: 100 KB · 200 KB · 500 KB · 1 MB · 2 MB · all targets.