Back to Updates

Compress a PDF: Common Errors and Fixes

A
Admin
2026-11-15
Last updated 2026-11-15

Quick answer

Four things go wrong with this tool, and each has one cause. A bigger file means the input had real text and did not need compressing. Unreadable small print means quality or resolution went too low. A refusal means the file broke the 25 MB or 50 page limit. A blank page means a page failed to draw, and the fix is a fresh run from the original. None of these is a mystery once you know which lever produced it.

What this article covers

The symptoms, the cause behind each, and the fix, in the order they are likely to appear. It also covers the two errors that are not the tool's fault — an encrypted PDF and a file that is not really a PDF — and what to do instead. Read the beginner's guide if you have not used the tool yet.

What it actually does

Because the tool redraws every page as a JPEG, the failure modes are the failure modes of image encoding, not of document conversion.

  • Size is driven by pixels times quality. More pages, higher resolution and higher quality all push it up.
  • Readability is driven by the same two numbers in the other direction.
  • Drawing is driven by page count and file size, because every page is rasterised in the tab.
  • Text is not preserved at all, so "the text is gone" is expected, not a bug.

Step by step

  1. Reproduce with the original file, not with a previously compressed copy.
  2. Note the produced size the page prints next to the download.
  3. Identify which symptom you have from the list below.
  4. Change one setting and run again.
  5. If the run is refused, split the document and compress the parts.

What to check after download

  • The produced size. If it is larger than the input, stop: this file is the wrong shape for the tool.
  • The smallest type. If it is unreadable at 100 per cent, quality was too low.
  • Every page present. Skim the thumbnails; a blank page is easy to miss.
  • The first and last page. Scanners often add or drop a blank at the ends.
  • A second viewer. If only one application opens the result, keep the original.

Fixes by symptom

The file got bigger. The input was born digital, with a real text layer. Re-encoding text as a picture costs more than the text itself. Fix: do not compress this file; send the original, or export a smaller PDF from the application that produced it.

The small print is unreadable. Quality or resolution is too low. Fix: raise quality first, in steps of 15, then resolution in steps of 30 dpi. A page of 9-point type needs more than a page of headings.

The document was refused. It is over 25 MB or over 50 pages. Fix: split it with a PDF splitter and compress the parts, or compress only the pages that need it if your source allows a page selection.

A page is blank. A page failed to draw, usually on a very large or unusual page. Fix: run again from the original; if it recurs, split that page out and handle it separately.

The PDF is encrypted. The tool cannot draw pages it cannot read. Fix: remove the password in the application that owns the file, then compress.

The file is not a PDF. Some downloads are HTML pages with a .pdf name. Fix: open it in a browser; if it shows a web page, download the real file.

A second compression made it worse. Each pass re-encodes an already encoded image. Fix: always start from the original.

Colour turned grey. Grayscale was left at yes from an earlier run. Fix: check the switch on every run; it is the one setting a preset cannot guess from the file.

The result is smaller but the recipient says it is blurred. The file was reduced to fit a limit and then displayed larger than it was rendered. Fix: raise the resolution for the size it will be read at, and keep quality where it was. A page meant to be zoomed needs more pixels than one meant to be skimmed.

Common beginner mistakes

Diagnosing a size problem by lowering quality when the real issue is that the input did not need the tool. The first question is always "does this PDF have selectable text?".

Re-running on the output instead of the input, which compounds loss twice.

Assuming a radius of pages is fine because the first page looked good. Check the busiest page, not the cover.

Fixing a size problem by retrying the same settings. If the produced size did not change, nothing in the pipeline changed either; the lever you touched was not the binding one.

Reading the produced size before the render finishes. The number appears next to the file, and it is the only measurement of this particular pass — a size seen on a previous document says nothing about this one.

When this tool is the right choice

When the page is a picture and the destination has a size limit. If any symptom above is "the text is gone", that is not an error — it is the tool working as designed, and the right fix is a different tool.

Related tools

If you need the images rather than the PDF, PDF Converter renders pages as JPEGs with its own resolution control. If the problem is too many separate files rather than one too-large file, Merge PDF Files is the tool that reduces the count.

Use this tool

Reproduce the problem from the original on Compress a PDF, change one setting, and read the produced size before downloading.

Next step

If none of the fixes above match, read best settings and start from a preset for your destination instead of from the defaults.

Tags