When to use Compress PDF

Use this for a scan when a smaller reading copy is worth a quality tradeoff. It rasterizes pages and removes text layers; other optimization approaches may preserve text, and a smaller output is not guaranteed.

How to Compress a PDF

Select a PDF — the tool checks whether it is a scan or a text document.
Choose a quality level.
Re-encode the pages locally in your browser.
Compare the before and after size, then download if the result is worth keeping.

What this tool changes

Replaces each page with a re-encoded image of that page at a lower resolution.

What it preserves

Page dimensions and page order, so the document still prints at its original size.

Good to know

This is lossy. Text becomes part of an image, so search, selection, copying, and screen-reader access stop working.
On a text-based PDF the output is usually larger than the input, not smaller.
Any existing OCR text layer is discarded — run OCR again on the compressed copy if you need it searchable.
Very large pages can exhaust the memory available to a browser tab.

This helps scans, and usually hurts text documents

This tool re-renders every page as an image at a chosen resolution and quality. It can reduce large scans, but it is not the only PDF optimization method and it discards selectable text, so compare the saved result before accepting that tradeoff.

For a document containing real text it is the wrong operation. Your text becomes pictures of text: search stops working, selection and copying stop working, screen readers get nothing, and the file frequently ends up larger, because characters are cheap to store and photographs of characters are not.

The tool checks which kind of document you have before you start and says so. If it warns you that the document contains real text, believe it — the honest answer in that case is usually to send fewer pages or re-export from the source, covered in PDF too large to email.

Choosing a quality level

Three presets trade sharpness against size. Higher quality keeps the most detail and saves the least; Balanced is the sensible default for ordinary printed text; Smallest file pushes hardest and is where small print starts to suffer.

Pick by what the document has to survive. A scan you only need to read on screen tolerates the smallest setting comfortably. A scan containing small print, figures, tables of numbers, or anything you may need to enlarge deserves a higher one — the detail discarded is not recoverable.

If the result is not small enough at Balanced, try Smallest before concluding compression has failed. If Smallest is still too big, the file is probably not the problem: a two-hundred-page scan is simply a lot of pages, and splitting or sending fewer of them is the real answer.

Check the result before you rely on it

Compression is not reversible. The discarded detail is gone, and any text layer the file had is gone with it — which is why the original is left untouched and worth keeping until you have looked at the output.

Look at a page or two, particularly the smallest print and any figures. Reading comfortably on screen is not the test if the document is going to be printed or read by someone with weaker eyesight than yours.

If a document comes out bigger than it went in, that is reported rather than hidden, and it is the expected result for a text-based PDF. Keep the original in that case: it is both smaller and still searchable. And if a compressed scan also needs to be searchable, run OCR afterwards — compressing after recognition throws the recognition away.

ZIP savings depend on the original PDF

A PDF with already compressed image streams may gain little from ZIP. Uncompressed or repetitive content can shrink, so measure the actual archive instead of applying one fixture result to every file.

That measurement took two attempts, and the first one was wrong in an instructive way. Run against a document whose pages were generated from a formula, zipping reported a 97.5% saving — which said everything about the test data being trivially compressible and nothing about real PDFs. Re-run with incompressible content, the saving was nothing. The full method is on the measurements page.

What does work depends on why the file is heavy. If it is a scan, the weight is in the images and re-encoding them is the lever this tool pulls. If it is a text document, there is very little to recover and compression will mostly cost you image quality for nothing. And if a single page is carrying the weight, removing it frees essentially its entire size — where cropping that same page frees nothing at all.

Frequently asked questions

Why did my PDF get bigger instead of smaller?: Because it was a text document rather than a scan. Text is stored as characters, which are extremely compact; replacing them with photographs of characters costs far more space than it saves. Keep the original — it is both smaller and still searchable.
Will the text still be searchable afterwards?: No. Compression rasterises each page, so any text becomes part of an image. If you need a searchable copy, run OCR on the compressed file to add a new text layer.
Is my file uploaded to compress it?: No. Pages are rendered and re-encoded in your browser, and the rebuilt document is assembled on your device.
Should I zip the PDF instead?: Try it if an archive is acceptable to the recipient, then compare sizes. Already compressed streams may shrink little, while repetitive or uncompressed data can shrink substantially.
Why did compressing my text document barely help?: Because there was little to recover. Characters and vector shapes are already compact; the savings live in images. Compression helps scans and photographs, and mostly costs quality for nothing on a text file.