Compress PDF

Because the thing making it big is almost certainly a picture, and tidying up the file structure doesn't touch pictures. A 40MB PDF is usually a scan, or a report somebody dropped phone photos into at full resolution. The text in it weighs almost nothing.

That's why there are two methods here and why they give such different results. Picking the wrong one either does nothing or ruins the document.

Tidy up

This rewrites the file's internal bookkeeping, packs it more efficiently and drops anything nothing points at any more. Deleted pages leave debris behind, and editors often keep old versions of objects around.

It's lossless. Every pixel and every character comes out identical. Nothing can go wrong.

It also usually saves very little. On a PDF exported cleanly from Word you might see five percent. On one that's been through several rounds of editing, sometimes thirty. On a scan, close to nothing, because a scan is one enormous picture per page and that picture is already compressed.

Try this first. If it gets you under the limit, stop.

Flatten to pictures

This renders every page as a picture and rebuilds the document around those. That's where the large savings live, and it's destructive in a way worth understanding before you press it.

Afterwards the text is not text. Nobody can select it, copy it, or search it. Ctrl+F finds nothing. A screen reader finds nothing, which means anyone using one can't read the document at all. Links stop working. Form fields disappear.

For a scan, none of that matters, because a scan already was pictures and had none of those things to lose.

For a contract, a CV, or anything somebody needs to quote from, it matters a great deal. A CV flattened to pictures is invisible to the software most employers screen with.

The two dials

Resolution decides how many pixels each page gets. 72 dpi is a screen, 150 is comfortable reading, 300 is proper print. Most documents that only need to be looked at are fine at 150. Going from 300 to 150 cuts the file to roughly a quarter, because you're halving in both directions.

Quality is the JPEG setting. Below about 60 percent you start seeing blocky patches around text edges, and text is exactly where JPEG artefacts show worst. 70 to 80 is the useful range.

If the result is still too big, drop the resolution before you drop the quality. A smaller sharp page reads better than a large mushy one.

Why a scan is so big

A page of text stored as text is a few kilobytes. The same page photographed is a million pixels or more, and no compression makes that as small as the letters it depicts.

If your scanner offers black and white for text documents, use it. A page scanned in colour carries three values per pixel to represent something that's only ever black or white. Switching modes often cuts a scan by eighty percent before any tool touches it.

If you need the words back out of a scan, that needs recognition rather than compression. Read the text off it instead.

What we can't do here

We can't recompress the images inside a PDF while leaving the text as text. That's what the good desktop tools do, and it's the ideal answer: shrink the photographs, keep everything else intact. Doing it properly means unpicking the PDF's internal image streams, handling every colour space and every compression type they might use, and putting them back without breaking the references. That's a big piece of work and doing it badly produces corrupt files.

So the honest position is two blunt instruments rather than one sharp one: lossless and modest, or drastic and destructive. If your PDF is mostly text with a few heavy images and neither option suits, a desktop tool will beat this.

Before you send it

Check the result before you attach it. Open it, search for a word you know is in it, and see whether it's found. If it isn't, you flattened something that shouldn't have been flattened, and the original is still sitting where you left it.