PDF compression is the most over-promised feature on the internet, so this guide starts with the truth: how much a PDF can shrink depends almost entirely on what is inside it, and the honest range runs from “half the size” to “barely anything”. Knowing which file you are holding, before you compress, saves the disappointment and points you at the fix that will actually work. Our free compress PDF tool does the lossless part in your browser, and this guide explains the rest.
In this guide
Where a PDF’s weight actually lives
Open a heavy PDF and the weight is almost never the text. A page of text with an embedded font costs a few kilobytes; images cost megabytes. The arithmetic is brutal: a single A4 page scanned at 300 dpi is about 2480 × 3508 pixels, roughly 26 MB uncompressed, and even as a decent JPEG it lands in the hundreds of kilobytes to low megabytes. A 20-page scan or a photo-rich report is therefore an image album wearing a PDF coat, and its size is decided by image resolution and image compression, nothing else. The remaining weight categories are embedded fonts (modest), duplicated or orphaned objects left behind by editors (variable), and metadata (small but interesting, as the PDF pillar notes).
Lossless compression: what the browser tool does
Lossless compression rewrites the file’s structure without touching any content. Concretely: packing the object catalog into compressed object streams, removing unused and duplicate objects, and stripping removable metadata. Every page renders identically afterward, byte-for-byte the same text and images, which is exactly what “lossless” promises. The wins depend on how wastefully the file was written: documents produced by simple generators and some export pipelines carry a lot of structural slack, and in our testing files like that can shrink by around half. A file that was already written tightly has little slack to reclaim, and lossless compression will say so by barely moving. That honesty is a feature: if the structure is clean, the remaining weight is content, and content reduction is the next section’s job.
Lossy compression: where the big wins hide
When a PDF must get dramatically smaller and it is image-heavy, something real has to be given up: resolution or image quality. Lossy PDF compression re-encodes the embedded images, downsampling 300 dpi scans to 150 or 100 dpi and recompressing JPEGs harder. The results can be spectacular, a 40 MB scan dropping to 4, and the cost is permanent: zoomed text gets softer, fine print can smear, and there is no undo. Two rules keep it safe. First, match the dpi to the destination: 150 dpi is fine for screen reading and most email, 300 for print. Second, keep the original. Lossy work of this kind is heavier than a browser tab should carry for large files, which is why our tool stays lossless and this guide tells you plainly when desktop software is the right call.
Honest expectations by file type
| What the PDF is | Lossless result | If you need more |
|---|---|---|
| Text-only export (invoice, essay) | Often already small; modest shrink | It is already light; check it is not carrying unused fonts |
| Generator output with structural slack | Large shrink, sometimes around half | Rarely needed |
| Office export with photos | Modest shrink | Re-export with lower image quality at the source |
| Scanned document | Small shrink; the images are the file | Lossy downsampling, or re-scan at sane settings |
The best compression happens before the PDF exists
The cheapest megabytes are the ones never written. Scanning at 300 dpi color for a black-and-white text document is the classic self-inflicted heavy file; 150 to 200 dpi grayscale reads identically on screen at a fraction of the size, and true black-and-white scanning is smaller still. Office exports usually have an image-quality setting hiding in the save dialog, and one notch down is rarely visible. And when a “PDF too large to email” is really a pages problem, ten signed pages inside a 60-page packet, the right tool is not compression at all but a split, covered in the merge and split guide.
Frequently asked questions
Why did my PDF only shrink 3%?
Because it was written tightly and its weight is content, almost certainly images. That 3% was the real structural slack. Getting further requires lossy image reduction or fixing the source, not running the same compression twice.
Does compressing twice help?
Lossless: no, the slack is gone after the first pass. Lossy: it removes more each time and damages quality each time, like photocopying a photocopy. Decide the target size once and compress once.
Will compression break the text or links?
Lossless compression preserves everything functional: text stays selectable, links keep working, pages render identically. Lossy image recompression also leaves real text intact, since it only touches images; text that degrades was a scan, meaning it was an image all along.
Is there a “right” size for an emailed PDF?
Under 10 MB passes essentially every mail system; under 25 MB passes most. If a document cannot reasonably get there, share a link instead of an attachment; no amount of compression makes a 200-page photo archive into an email-sized object honestly.