Key takeaways
- Text-only PDFs rarely compress, because characters take a few kilobytes while a picture of the same page takes far more space than the text did.
- Compressing an already compressed PDF saves very little and adds blur, so always rerun the original file at a stronger compression level instead.
- Halving image resolution leaves a quarter of the pixels, which is why downsampling a 600 DPI scan to 300 or 150 DPI shrinks it dramatically.
- Re-exporting from Word, Preview or a scanner at sensible settings often beats compression, because text stays selectable while only the images get smaller.
A PDF usually won't compress any smaller for one of three reasons: there is nothing heavy inside it to shrink, it has already been compressed, or the compressor's method does not suit what the file contains. Text and vector graphics are already compact, and images that were squeezed once have little left to give. The fix is to find out what is taking up space, then either export the file again from its source or downsample the images inside it.
Diagnose it in two minutes
Open the PDF and try to highlight a sentence. Then look at the size and page count. Those three facts point to the cause in most cases.
| What you see | Likely cause | What works |
|---|---|---|
| Text highlights, file is small, compressor says it is already compact | Nothing heavy to shrink | Send it as it is |
| Text highlights, file is large, few pages | Large photos or graphics inside | Re-export with smaller pictures, or downsample images only |
| Nothing highlights, many pages | A scan saved at high resolution | Compress at a stronger level, or rescan |
| A second compression saves almost nothing | Already compressed once | Go back to the original file |
| Hundreds of pages, every one a scan | Too many images for the target | Remove or split pages as well |
To see which images are heavy, Extract Images from PDF pulls out each embedded picture at its stored pixel size. Download the large ones and check their dimensions: a 4,000-pixel photo on a page that only needs 1,000 pixels is your answer.
Why some compressors make files bigger
Our Compress PDF tool works by redrawing each page as a JPEG image, then rebuilding the document. That is very effective on scans, where every page is already a big picture. It is the wrong method for a page of plain text. Text in a PDF is stored as characters and font instructions, which take a few kilobytes. A picture of that same page takes far more.
So when you compress a text-only PDF, the result would often be larger than the original. The tool checks for this and tells you the file is already compact, with no download, rather than hand you a worse copy. That message is useful information: your file has no images worth shrinking.
Compressing twice does not work
JPEG compression throws detail away each time. Run a compressed file through again and the second pass has very little left to remove, so the size barely changes while the blur gets worse. If the first result was too big, always start again from the original at a stronger level.
The same applies to scans that were already saved as small JPEGs by a phone app. They are close to their floor, and squeezing harder mostly adds blocky artefacts around the letters.
Downsampling, explained
Downsampling means reducing the number of pixels in an image. Because file size tracks pixel count, it is the single biggest lever on a heavy PDF.
| Scan resolution | Pixels on a US Letter page |
|---|---|
| 600 DPI | about 33.7 million |
| 300 DPI | about 8.4 million |
| 150 DPI | about 2.1 million |
Halving the resolution leaves a quarter of the pixels. That is why a 600 DPI scan is so heavy, and why 300 DPI is plenty for documents.
There are two ways to downsample. The first redraws the whole page as an image, which is what our tool does at about 144, 108 or 72 DPI depending on the level. It is simple and strong, but the text stops being selectable. The second shrinks only the embedded images and leaves text as text. Desktop tools do this second kind. Ghostscript is a free command-line option, and this command downsamples images to about 150 DPI while keeping text selectable:
gs -sDEVICE=pdfwrite -dPDFSETTINGS=/ebook -o smaller.pdf original.pdf
The /screen preset goes lower, to about 72 DPI, and /printer keeps about 300 DPI.
Re-export instead of compressing
If you have the source document, exporting it again is often better than any compressor, because the text stays real and the images can be shrunk properly.
- Microsoft Word. Use Picture Format > Compress Pictures to lower the resolution of photos in the document. Then save as PDF and, on Windows, in the Save As dialog, choose Minimum size under Optimize for.
- Mac Preview. Choose File > Export, pick PDF, and select the Reduce File Size Quartz filter. It is aggressive, so check that photos still look acceptable.
- Scanner apps. Scan documents at 300 DPI, and use grayscale or black and white for plain text pages. A colour scan stores three colour channels per pixel instead of one, which adds weight for no benefit on a black-and-white letter.
- Photos you place yourself. Resize them before building the PDF. Our explainer on why PDFs get so large covers this habit in more detail.
If it still will not fit
Some files are simply big: a 300-page scanned archive will not fit a 2 MB limit at any readable quality. At that point, change the file rather than the settings. Remove pages the recipient did not ask for, split the document into parts, or share a cloud link instead of an attachment.
One more caution before you settle for an image-based compressor. The compressed copy keeps the look of each page, but links, bookmarks, form fields and any attached files are not carried over, and the text can no longer be copied or searched. Keep the original, convert or edit it first, and compress last. For the standard step-by-step method, see our guide to reducing PDF file size for email and uploads.
Frequently asked questions
Most often the PDF is mostly text or vector graphics, which are already compact, so there is nothing heavy to shrink. Other causes are a file that was compressed before, or images that are already low resolution. Check whether the text highlights and look at the page count to see which case you have.
Compressors that redraw pages as images can produce a larger file from a text-only PDF, because a picture of a page takes more space than its text. Our Compress PDF tool checks for this and shows an already compact message instead of producing a larger file, so you never download a worse copy.
Downsampling reduces the number of pixels in the images inside a PDF, for example from 300 DPI to 150 DPI. Halving the resolution leaves a quarter of the pixels, so the file shrinks sharply. Some tools downsample only the embedded images and keep text selectable, while others redraw whole pages as images.
Yes, but you need a method that shrinks only the images. Exporting again from Word with smaller pictures, Mac Preview with the Reduce File Size filter, or Ghostscript with the ebook preset all keep text as text. Image-based compressors, including ours, turn each page into a picture, so the text stops being selectable.
No. Each JPEG pass throws away detail, and the second pass has little left to remove, so the size barely drops while the blur increases. If the first result is still too large, go back to the original file and compress it once at a stronger level, or remove pages you do not need.