You've got a PDF. It's 35MB. The email portal caps attachments at 10MB. You try to compress it with some random online tool, it comes back as 34.8MB, and now you've wasted five minutes and learned nothing.
This happens constantly, and it happens because most people skip straight to "how do I fix it" without understanding what's actually making the file large. Once you know that, the fix becomes obvious.
Why your PDF is so heavy
A PDF is a container. It can hold text, images, fonts, vector graphics, form fields, annotations, color profiles, and a pile of invisible metadata that accumulates during the creation process. The weight of the file depends almost entirely on what's inside it and how efficiently it's been stored.
In practice, images are almost always the main offender. When you export a document from Word, InDesign, or Illustrator, images get embedded at whatever resolution they were in the source, often 300 DPI or higher because that's what you'd need for print. A single full-page photo at print resolution can easily be 8 to 15MB on its own. For a PDF that will only ever be read on a screen, that's completely wasted space. Screens don't need more than 96 to 150 DPI. Nobody zooms in to a PDF far enough to see the difference.
Fonts are the second contributor, though less dramatic. To ensure the document looks identical on any machine, font files get embedded directly into the PDF. A single font family can add 200 to 500KB. If your document uses five different fonts, that adds up to a couple megabytes before you've included a single image.
Then there's the invisible bloat: deleted content that wasn't actually purged, form field history, embedded preview thumbnails, ICC color profiles, JavaScript leftover from interactive forms, and metadata from every tool that touched the file. None of this is visible to the reader, but it's all taking up space.
Finally, some PDF generators simply write inefficient file structures. The data is all there, but packed poorly, like a bag you stuffed in a hurry instead of folding things properly.
What actually works
Resampling the images is where you get real results. A tool that genuinely compresses a PDF will re-encode embedded images at a lower DPI, reducing 300 DPI photos to 150 or 96 DPI. This is where 60 to 80% size reductions come from. The text stays sharp because text in a PDF is stored as vectors, not pixels. Your document will look identical on screen.
If you're not seeing those kinds of reductions, the tool probably isn't doing this step. Some online compressors just clean up the file structure and strip some metadata, which trims maybe 5 to 15%. That's not nothing, but it won't save you when you're trying to get from 35MB to under 10MB.
Re-exporting from the source is the cleanest option when it's available. If you have the original Word or Google Docs file, export it using "Save as PDF" rather than "Print to PDF." They're not the same thing. Print to PDF rasterizes everything, turning vector text into pixel images, which can actually make the file larger and destroys text selectability. Native export handles fonts and images much more efficiently.
Stripping unnecessary elements helps at the margins. If you have a PDF editing tool, you can remove embedded thumbnails, color profiles, and metadata. It's not where the big wins are, but if you've already compressed the images and you're still slightly over a file size limit, this can push you over the line.
Converting scanned documents to grayscale is worth mentioning because people overlook it. If you have a scanned form or a photocopied contract that's inherently black-and-white, there's no reason for the embedded images to be stored in RGB color. Converting to grayscale can cut image sizes roughly in half.
What doesn't work
Re-compressing a PDF that's already been compressed doesn't really work. If someone sent you a PDF that's already been through a compressor, running it through again will save almost nothing and can degrade image quality. The first compression already discarded most of the redundant data.
Printing to PDF as a compression trick makes things worse, not better, for reasons already covered above.
And there are tools online that claim to compress PDFs but just rewrite the file header or change the metadata. If your 35MB file comes back as 34.9MB after "compression," nothing happened. A genuine compressor should be getting you at least 30 to 40% on a typical document, often much more if images are the primary content.
A quick note on quality loss
For screen-only PDFs, compression to 150 DPI is essentially invisible. You'd have to zoom in past any normal reading level to notice a difference. At 96 DPI, text in screenshots or photos might look slightly softer at extreme zoom, but at normal reading size on any monitor, it holds up fine. The documents that show obvious quality loss after compression are usually ones with very small text embedded in images, like scanned forms where the text was already marginal to begin with. In those cases, you can stay at 150 DPI and still get solid size reductions.
If a PDF is mostly charts, graphs, or illustrations with a lot of fine detail, be more conservative with compression settings. If it's mostly text with a few photos, compress aggressively. The content type matters more than a fixed DPI rule.
One thing worth checking first
Before you compress anything, open the PDF and check the file properties. Most PDF viewers will show you the page count, whether the file is linearized, and sometimes the embedded font list. If your 35MB file has 6 pages and 12 embedded fonts but no images, image compression won't help much. You're dealing with a font and structure problem instead. Re-exporting from the source is the better path in that case.
If the file has images, even a few, that's almost certainly where the weight is coming from.