Compress PDF

Make a PDF smaller by storing the same document more efficiently. The work happens in this browser, on this computer: the file is not uploaded, because burrow has no server that could receive it.

Nothing is re-encoded. No image is resampled, no font is stripped, no page is redrawn. Every page's contents come through unaltered — only the way the document stores them changes. A scan looks exactly as it did, because burrow will not make a file smaller by making it look worse.

This tool needs JavaScript

The compressing is done by code running in your browser, so with JavaScript switched off there is nothing to do the work. There is no server-side version to fall back to — that is the whole design, not a missing feature.

With JavaScript on, this page lets you choose a PDF, compress it, and download the result — or be told plainly that it was already efficiently stored. Nothing is sent anywhere.

What this will and will not shrink

burrow makes a PDF smaller by changing how it stores its structure — packing the thousands of small pieces of bookkeeping a PDF carries into shared blocks, and dropping whatever nothing points at any more. It never touches your pages.

So how much a file loses depends almost entirely on what it is made of:

Forms, reports and anything generated by software are assembled from many small parts, and those are exactly what this packs together. These can lose a great deal.
Scans and photograph-heavy documents usually lose almost nothing. A scan is a stack of images. The image data is already compressed — that is what makes it a JPEG rather than a bitmap — and there is barely any structure around it to reorganise. The only way to make such a file meaningfully smaller is to re-encode the pictures, which means throwing away detail, and burrow does not do that.
A file that has already been through a tool like this one is already stored the efficient way, so there is nothing left to do. burrow will tell you so rather than hand you back a copy of the same size.

This is said here, before you choose anything, because the alternative is letting you wait for a number that looks like a broken tool. A file that does not shrink is not a failure and burrow does not report it as one — it is a fact about the file, and usually a good one.

What you are shown

The size before, the size after, and the percentage — measured, not estimated. If the saving is too small to be worth replacing your file for, burrow says that instead of presenting a rounding error as a result. The compressed document is still offered either way; the page reports what it found and leaves the choice with you.

If burrow's version came out no smaller than yours, it keeps yours and tells you both numbers. Nothing is hidden behind a cheerful message.

What happens to large files

Everything is held in this browser tab's memory, so the practical limit is what your device can spare rather than an upload size. burrow refuses a file larger than 512 MB before it opens it, and a document of more than 10,000 pages before it compresses any of it — so an oversized job is refused rather than failing part way through.

What that does not save you from, said here rather than discovered: compressing is a single long operation inside the PDF engine, and unlike turning or reordering pages there is no point part way through at which it can be interrupted cleanly. A very large document can make the tab slow while it runs, and on a phone the browser may stop the tab entirely. Stop ends it by discarding the engine, which is the only thing that can.

What is kept, and what is not

Everything. Bookmarks, attachments, form fields, annotations, metadata and the page contents all come through unchanged. Compressing rewrites how the document is stored and nothing about what it says, and burrow's tests check the claim rather than assuming it: every marked object the document carried must still be there afterwards, and each page's drawing instructions must come through unaltered.

That is worth stating plainly because many tools called "compress PDF" do the opposite — they shrink a file by re-encoding its images at lower quality, which is why a scan run through one comes back blurry. burrow's answer to a scan is to tell you it cannot help much, rather than to help by damaging it.

A file that is encrypted, damaged, or not a PDF is refused. So is a result burrow produced but could not verify: after compressing, it reopens the document it just wrote and checks that every page is still there and still turned the same way. That check is deliberately narrow — it is what a fresh engine can confirm about a document in the moment it is produced, and the promise that your pages are untouched rests on burrow never re-encoding them in the first place, which the tests above check. If the check fails, you get the refusal rather than the document.