CloudlessKit / Guides

How to make a PDF smaller (and why it got so big)

Updated July 22, 2026

A 20-page contract that's mostly typed text should weigh a few hundred kilobytes, not fifteen megabytes. If your PDF is way bigger than that, the extra weight almost always comes from images baked into the file, not from the words on the page. Figure out which kind of image before you try to fix it, because a scanned document and a report full of photos need different treatment.

Why this file is so big in the first place

Open the PDF and try to select a line of text with your cursor. If nothing highlights and the whole page behaves like one flat picture, you're looking at a scan: every page is a photo of a paper document, usually saved at a resolution meant for printing, not for screens. A single scanned page can easily run 1 to 3 MB on its own, so a 20-page scanned statement adds up fast. This is the single most common reason a PDF is huge.

If you can select and search the text normally, the bloat is coming from somewhere else. Two usual suspects: embedded photos (a report with a dozen full-resolution product shots or screenshots dropped in at their original camera size), or embedded fonts. Some PDFs bundle the complete font file for every typeface used, sometimes several weights of it, and that alone can add a few megabytes even in a document with almost no images. There's a fourth possibility worth naming: sometimes a PDF really is just tight already, because it's mostly vector text and simple line art, and no tool is going to shrink it much further. Squeezing an already-lean file rarely buys you anything.

The 5 MB portal problem: what to do right now

If a government or bank portal is telling you the cap is 5 MB and your scanned file is 40 MB, don't panic and don't start emailing IT. First check whether the portal accepts multiple files per upload field, or a follow-up email with attachments. If it does, splitting a long scan into two or three smaller PDFs and uploading each in its own slot is often faster and more reliable than trying to compress your way under the limit, especially if the portal only needs certain pages (the signature page, not the full 60-page appendix).

If it truly needs to be one file, compression is your move, and for a scan it usually works dramatically well. Dropping a 40 MB stack of scanned pages down to the resolution a screen actually needs (rather than the resolution a laser printer wants) can cut the file to a tenth of its size or less, because you're removing pixels the reviewer was never going to see anyway.

Downsampling a scan versus recompressing a photo

These sound like the same problem but they're not quite. A scanned page is usually stored at a resolution far higher than a screen needs, often 300 dots per inch or more, meant for a printer. Cutting that down to something screen-appropriate is where almost all the size savings live, and a human reading it on a laptop or phone won't notice much difference. A photo embedded in a report is different: it's already sized for viewing, so the win there comes mostly from re-encoding it at a lower JPEG quality, trading a bit of fine detail (skin texture, sky gradients, fine text inside the photo) for a smaller file.

The PDF Toolkit's Compress PDF tool treats both cases the same way under the hood: it re-renders every page as a JPEG at one of three settings (Light, Balanced, or Strong), each a step down in resolution and quality, and rebuilds the PDF from those images, entirely on your device with nothing uploaded. That approach fits the scan-heavy and photo-heavy cases described above. If you run it on a file that's already efficient, it will tell you rather than making the file bigger for no reason.

What you give up when you compress

Here's the tradeoff, plainly: once a page has been through an image-based compressor, the text on it is no longer selectable or searchable, because the whole page is now one picture instead of text characters plus a picture. That's a real cost, not a minor footnote. If you need to keep a document searchable, either compress a copy and keep the original for reference, or run a searchable-text pass on the compressed file afterward, since compressing rebuilds every page as a plain image and erases any text layer that was there before, including one added by OCR. Fine print and dense tables also survive compression worse than plain paragraphs; if a setting turns your smallest font blurry, back off one level and check whether the size savings even matter that much.

Fonts, though, are a different story.

If font data is the real cause of the bloat, compressing pages as images can still shrink the file, since the whole page becomes a JPEG and the old font data goes away with it, but you're trading searchable, sharp text for a fixed image of that text just to solve a font-size problem. That's usually a worse trade than it sounds. If keeping real text matters and fonts are the culprit, your best bet is going back to whoever produced the PDF and asking for a version exported with fonts subset (only the characters actually used, not the whole typeface). Desktop PDF editing software can generally do this; most quick online tools can't.

When splitting beats compressing

Compression isn't always the right tool. If an email provider caps attachments somewhere around 20 to 25 MB and your file is a 200-page scanned case file, cutting resolution alone may not get you there without wrecking legibility, and splitting the document into two or three parts along natural breaks (by date, by section, by exhibit) keeps every page fully readable. It's also just easier on the person receiving it: nobody wants to scroll through 200 pages to find exhibit C when it could have arrived as its own file. The PDF Toolkit's Split tool can break a long PDF into smaller pieces, and Merge can combine several files back into one if you need to reassemble things a different way later, and neither one requires re-encoding a single pixel.

One more honest note: on a Mac, the built-in Preview app has a quick way to shrink a PDF's file size (worth a try before reaching for another tool), and it can help with a lightly bloated file in a pinch, though the result is a bit of a black box and you can't control the quality level the way you can with a dedicated compressor. Windows doesn't ship anything comparable, so you're reaching for a separate tool either way. Whichever tool you use, keep the original file until you've confirmed the compressed version still opens cleanly and reads clearly on screen.