Why compressing a PDF sometimes makes it bigger

If a compressed PDF is larger than the original, keep the original. A compressor may be rebuilding the document in a less efficient form, even when it makes the pages look roughly the same.

The ToolsNow PDF compressor renders each page as a JPEG image and puts those images into a new PDF. That can help with large scans. It can also enlarge a compact report and remove its searchable text.

Decide whether a picture-only copy will do

Before changing settings, check what the recipient needs. If they need to search, copy text, fill in fields or use the document’s accessibility structure, keep a version that preserves those features.

Your document	A useful next step
Large scans, needed only as readable page images	Try compression on a copy and compare the download
Report with selectable text and a few large photos	Re-export from the source with smaller images, if available
Scan that already has searchable OCR text	Use an optimiser that preserves its text layer
Form, signed document or accessible publication	Use a workflow that preserves and verifies those features
Already-small PDF	Keep it unless a specific limit requires a smaller version

Why the same settings give different results

A PDF can contain text, vector drawings, photographs, scanned pages or a mixture. Text and reusable fonts can describe a page more compactly than a picture of every letter. Turning that page into an image can add data while losing the text’s ability to stay sharp at different zoom levels.

A high-resolution scan may have room to shrink: fewer pixels or lower JPEG quality can reduce the stored image. But some scans already use efficient black-and-white compression. Replacing those with colour JPEGs can make them larger too.

There is no dependable percentage saving for either category. Page count and file size alone do not tell you what’s inside. A short report with one large photograph can outweigh a much longer text document.

Try a copy and compare it at reading size

1. Keep the original PDF and open a copy in the compressor.
2. Check Resolution and Image quality. The current defaults are 96 DPI and 80%; they are starting settings, not a quality guarantee.
3. Choose Compress PDF and compare the reported old and new sizes.
4. Open the downloaded file. Check the page count, small print, faint text, diagrams and any barcodes at the size you expect people to use.
5. If detail has suffered, try a higher resolution or quality using the original. If the result is larger or less useful, keep the original.

The tool can download a result even when it reports No saving this time. The filename ending in -compressed.pdf does not mean that copy is smaller.

Lower resolution removes pixels. Lower JPEG quality changes how those pixels are encoded. Changing one setting at a time makes it easier to see which trade-off is acceptable. Repeatedly compressing an already-compressed copy adds another round of loss.

What happens to search, forms and signatures

This export contains page images rather than the original text objects. It does not retain the existing text layer, including OCR text behind a scan, or recreate the document’s tags, interactive forms, links or digital signatures. A signature may remain visible as a picture without retaining its cryptographic verification.

Some viewers can recognise text in images themselves. If search works in one of those viewers, that does not establish that the exported PDF has a text layer other software can use.

Rendering also changes how detail looks, particularly when zooming or printing. A readable screen preview is not proof of an identical print.

Check for text before you start

Try selecting and copying a sentence, then search for a word you can see. The PDF to text extractor can also show the text it finds. Its scan warning means it found very little extractable text; it is not a complete diagnosis of the file.

If you have the source document, reducing oversized pictures there often gives you more control. Other PDF optimisers can work on embedded images, fonts and unused objects without rasterising every page. Check what a particular tool changes before choosing it for a searchable document.

Last reviewed 2 October 2026.