PDF Still Too Big After Compressing

You dropped the file into a compressor, waited, and got back something 4% smaller than what you put in. Or worse, slightly larger.

That is not a broken tool. It is the tool telling you something about the file, and there are only five realistic explanations. Each one has a different way forward, and three of them involve giving up on compression entirely and attacking the problem somewhere else.

Work out which wall you have hit

Cause 1: the file is already optimised

This is the most common and the least satisfying answer. A PDF exported properly from Word, generated by LaTeX, or produced by a modern publishing tool has already had its fonts subsetted, its objects deduplicated and its streams compressed. There is nothing left for a lossless optimiser to find.

How to confirm it: divide the file size by the page count. Under about 100KB per page on a text document means you are already at the efficient end. A lossless rebuild might find 5–30% by removing edit history and duplicate objects, and that is the ceiling.

What to do: stop compressing and start removing. Delete pages you do not need, remove embedded attachments, strip the high-resolution logo that appears on every page. If the file is 200 pages of text and genuinely needs to be smaller, compression is not the tool — content reduction is.

Cause 2: the tool only ran a lossless pass

Many free compressors default to lossless optimisation because it is safe. On a scanned document, safe means useless.

Lossless compression works by removing redundancy — duplicate fonts, unused objects, an inefficient cross-reference table. A scanned page is a photograph. A photograph of text contains exactly as much data as a photograph of a landscape, and there is nothing to deduplicate. You can run a 50MB scan through a lossless optimiser all day and save 2%.

How to confirm it: try to select a word in the document. If nothing highlights, the page is an image.

What to do: switch to downsampling, which re-encodes the page images at lower resolution. In our PDF compressor, that is the Downsample pages to images option — set 150 DPI and tick greyscale for a black-and-white document. This is lossy and irreversible, so keep the original until you have checked the result.

Expect 85–97% on a 600 DPI colour scan. That is the difference between the two modes, and it is why picking the right one matters more than picking the right tool.

Cause 3: the PDF is encrypted

If the tool threw an error, hung, or returned the file unchanged, check whether the document is protected. In Acrobat Reader, File → Properties → Security shows the permissions.

There are two kinds of protection and they behave differently:

• Open password. You need it to view the document at all. No tool can process the file without it.
• Permissions password (owner password). The document opens freely but editing, printing or content extraction is restricted. Compressors need to rewrite page content, so this blocks them too.

What to do: open the PDF with the password, then use Print → Save as PDF or your editor’s Save As to produce an unprotected copy. Compress that, then re-apply protection afterwards if the document needs it.

Two caveats worth stating plainly. Print-to-PDF rasterises text and destroys any searchable text layer, bookmarks and links — so prefer Save As from a proper editor where you have one. And removing protection from a document you do not own the rights to is a legal question, not a technical one.

Cause 4: the file got larger

This happens when a compressor rasterises vector content. Charts, diagrams, line art, maps and CAD exports are stored as mathematical paths — a complex diagram can be a few kilobytes. Convert it to a 300 DPI image and the same page becomes hundreds of kilobytes.

How to confirm it: zoom to 800%. Vector artwork stays perfectly sharp at any zoom; a rasterised version goes blocky.

What to do: do not downsample a vector-heavy document. Use lossless optimisation only, or re-export from the source application. If the file contains both vectors and photographs, use a tool that downsamples embedded images in place rather than re-rendering whole pages.

Cause 5: the target is unrealistic

A 40-page colour brochure is not going under 1MB while remaining recognisable. Some limits simply cannot be met by compression.

Four things that work when compression has run out:

Flatten the document

Form fields, hidden layers and transparency groups carry real weight — often 10–30% of a file that otherwise looks simple. Flattening merges all of it into fixed page content.

This is irreversible. Flattened form fields keep their values visually but can no longer be filled in, edited or read programmatically. If anyone downstream processes the form data, flattening breaks that permanently. Keep an unflattened master.

Split it

Most upload forms that accept one 5MB file will accept three files of 5MB. Splitting copies pages without re-encoding, so quality is completely untouched. It costs you nothing except the recipient receiving several files.

This is the only option in this guide with no quality penalty at all, which makes it the first thing to try when the limit is per-file rather than per-submission.

Re-scan at lower DPI

If the document came from paper and you still have the paper, re-scanning beats compressing. You avoid creating the excess data rather than throwing it away, and the result is visibly cleaner than a downsampled 600 DPI scan.

Set the scanner to 200 DPI, greyscale (or black and white for clean printed text), and medium quality. A clean page in black-and-white mode can come out under 50KB.

Re-export from the source document

The best option of all, when the source still exists. Re-exporting from Word, PowerPoint, InDesign or Pages with a screen-oriented preset gives a better result than any amount of after-the-fact compression, because:

• Images are downsampled once from the original rather than re-encoded from already-compressed versions.
• Text stays as text instead of being rasterised.
• Links, bookmarks and the document outline survive.
• Fonts get subsetted properly.

In Word: Save As → PDF → Options → Minimum size. Before that, select each image and use Picture Format → Compress Pictures with Delete cropped areas of pictures ticked — cropped-away image data is silently carried into the export otherwise, and on an image-heavy document that one checkbox can halve the file.

What not to do

• Do not run it through three different compressors in sequence. Each pass re-encodes already-degraded images. The artefacts compound visibly and the additional saving is marginal. Go back to the original and compress once with harder settings.
• Do not zip it. PDFs are already internally compressed. A zip saves a few percent and some upload forms reject zips outright.
• Do not assume a paid tool will succeed where a free one failed. If the file is already optimised, Acrobat Pro will report the same ceiling. The limit is in the file, not the software.
• Do not upload sensitive documents to unfamiliar online services while hunting for a tool that works. The documents people struggle to compress are overwhelmingly scans of personal paperwork.
• Do not delete the original at any point in this process. Every lossy step described here is one-way.

For a cause-by-cause breakdown of what makes PDFs large in the first place, see why is my PDF so large and what will actually shrink it. If you are working to a specific upload limit, compress a PDF to under 1MB or 200KB has the per-page budgeting.