Compress PDF to Reduce File Size

Work out why the file is large before you shrink it

Large PDFs are large for a short list of reasons, and only some of them are things a compressor can address. Open the document and try to select a sentence. If individual characters highlight, the page is real text and the weight is coming from somewhere else. If you get a blue rectangle across the whole page, the page is an image and you have already found the cause.

Beyond scans, three things routinely inflate a file. Fonts embedded in full rather than subset, so the entire typeface ships inside a document that uses eleven characters of it. Vector artwork carrying an enormous number of path points — maps, floor plans and CAD exports, where one page can outweigh a hundred pages of prose. And revision history: a PDF saved repeatedly by an editing application keeps its earlier states appended to the file, so what you are holding is the latest of several versions stored in one wrapper.

Only the first of those is an image problem. The other two explain the files that refuse to shrink no matter which preset is chosen.

What each preset is actually choosing

The three presets are not vague quality dials. Each one sets a target resolution for the images in the document, picked for a destination.

Printer retains enough resolution for a commercial press, which is why it is the mildest of the three. Around 300 dots per inch is the long-standing threshold below which printed photographs start to look soft.

Ebook aims at the middle ground — reading on a laptop or tablet, and printing on an office laser printer, where roughly half that resolution is indistinguishable on paper for most documents.

Screen is for files that will only ever be read on a display. A screen renders somewhere in the region of 72 to 96 pixels per inch, so detail beyond that is invisible on screen however much of it the file carries. That is exactly why Screen can be dramatic on a scan at no visible cost — right up until somebody prints it.

Pick by where the file is going, not by the number you would like to see when it comes back.

A worked example, and the limit that drives most of this

A 30-page scanned lease at around 40 MB is the standard case: thirty photographs of text, produced by a copier left on full colour for pages that are black ink on white paper. It will not attach to an email. The common mailbox ceiling sits near 25 MB — both Gmail and Outlook refuse attachments above roughly that size — and clearing it is the real goal behind most compression.

Run it once at Ebook, then inspect the result before reaching for anything stronger. Open the compressed file, zoom to 100%, and read the smallest text on the busiest page. If that is comfortable, stop; you are done. If the file is still too large, go back to the 40 MB original — never to the compressed copy — and run Screen on it instead.

One pass from the source beats two passes stacked on each other, every time.

When a smaller number is the wrong goal

Image compression is lossy and permanent, so some documents should not go through it at all.

Anything that will be printed larger than it was scanned — a plan, a poster, a certificate — needs the resolution it has. A page that looks perfectly acceptable on screen can fall apart on paper at size.

Anything submitted to a court, a registry or a regulator usually arrives with a stated specification attached, and meeting that specification matters considerably more than the file size does.

A scan is often better served by OCR PDF first. OCR adds a searchable text layer, which makes the document genuinely more useful, and you can still compress afterwards if it remains too large. Running it in that order also means the text layer is read from the sharper original rather than from a downsampled copy.

And if the file is large simply because it contains a great deal of material, compression is not the answer at all. Send only the part that is needed with Extract Pages, or divide it with Split PDF.