PDF stands for Portable Document Format — a format Adobe created in 1993 to let people share documents that look the same on every screen, every printer, every operating system. That consistency is PDF's superpower, but it also means PDFs carry a lot of data. A simple 5-page Word document might export as a 2 MB PDF. A 30-page report with photos? Easily 25 MB or more.
This guide explains what actually causes large PDFs, the difference between metadata cleanup and lossy compression, and exactly what a browser-based cleanup tool can and cannot do for your files.
What Actually Makes PDFs Large
Before you compress anything, it helps to understand what's eating up the bytes. Most people assume text is the problem. It almost never is.
Images are the number one contributor to PDF file size. A single full-page photograph at 300 DPI can easily be 5-8 MB on its own. If your document contains ten photos, you're looking at 50-80 MB before anything else factors in.
Here's what contributes to PDF size, roughly ranked by impact:
- Embedded images — Photos, screenshots, scanned pages, charts rendered as bitmaps. At 300 DPI with full color, these dominate file size.
- Full font sets — If your PDF embeds an entire typeface instead of just the glyphs actually used, that alone can add 200-500 KB per font.
- Uncompressed vector graphics — Complex vector illustrations, logos with transparency, and decorative elements stored without compression.
- Metadata and thumbnails — PDF thumbnails, document metadata, and XMP data add up, though typically only tens to hundreds of kilobytes.
- Duplicate or redundant content — Reused images, duplicate object definitions, and uncompressed streams that could be compressed.
The takeaway: if your PDF is large, start by looking at the images. Everything else is secondary.
Lossless vs Lossy Compression
These two terms describe fundamentally different philosophies, and understanding them is the key to choosing the right approach.
Lossless Compression
Lossless compression removes statistical redundancy from data. The information is encoded more efficiently, but absolutely nothing is discarded. When you decompress the file, you get back a byte-for-byte identical copy of the original.
In PDFs, the most common lossless algorithm is Flate compression, which uses the zlib library based on DEFLATE. It combines LZ77 compression with Huffman coding to find repeated patterns and encode them compactly. Every modern PDF tool uses Flate as its default compression for text streams and vector data.
When to use lossless: Always, as your starting point. There is zero quality trade-off.
Lossy Compression
Lossy compression permanently discards data that the algorithm calculates humans are unlikely to notice. In PDFs, lossy compression applies almost exclusively to images, most commonly through JPEG encoding. JPEG uses the Discrete Cosine Transform to convert image blocks into frequency data, then quantizes the high-frequency components that contribute fine detail.
You'll see blockiness in smooth gradients, ringing around sharp edges, and color banding in subtle transitions. At light compression (quality 85-95%), these artifacts are invisible to most people. At heavy compression (quality 20-40%), the degradation is obvious.
When to use lossy: When you need maximum size reduction and the document will be viewed on screen, not printed at high quality.
The critical point: lossy compression is irreversible. Always keep your original uncompressed PDF if there's any chance you'll need higher quality later.
What Our Compress PDF Tool Actually Does
It is worth being precise about what a browser-based tool can and cannot do, because "compression" covers very different things. ToolsKit Compress PDF removes hidden metadata and rewrites the file structure — it does not re-encode embedded images, so it never touches visual quality or selectable text.
Light (structure rewrite only)
This level keeps all metadata and only rewrites the file's internal object structure in a more compact form. Nothing is removed from the document.
Best for: When you want a structural cleanup but need the document's metadata left intact.
Standard (removes application info)
Standard clears the producer and creator fields, which record the software that generated the file. Title, author, and keywords are kept.
Best for: General sharing where you want to drop the generating-application details but keep the document's own descriptive fields.
Deep (strips all metadata)
Deep clears every metadata field — title, author, subject, keywords, producer, and creator — leaving no identifying document information attached. Because text and embedded images are copied through unchanged rather than re-encoded, there is no quality change at any level.
Best for: Publishing a document publicly or sending it to a new recipient when you do not want it carrying author names or internal tool details.
One honest limit worth knowing: because this tool never re-encodes embedded images, a file whose size comes mostly from scans or photos will shrink only modestly. The result page always shows the exact original size, new size, and percentage, so you can see the real outcome rather than a guess.
How to Compress Your PDF (Step by Step)
Here's exactly how to clean up a PDF's metadata using the ToolsKit Compress PDF tool:
- Open the tool. Navigate to the Compress PDF page in your browser. It works on desktop and mobile.
- Upload your file. Drag and drop your PDF into the upload area, or click to browse and select it. The file never leaves your device — processing happens entirely in your browser.
- Choose a cleanup level. Select Light to keep all metadata, Standard to drop the generating-application details, or Deep to strip every field including title and author.
- Run the cleanup. The tool removes the selected metadata and rewrites the file structure.
- Check the result and download. The result page shows the exact original size, the new size, and the percentage reduction. Save the cleaned file to your device.
The entire process takes seconds. No signup, no watermarks, no server upload.
Choosing a Cleanup Level by Use Case
Which level you pick depends on why you are cleaning the file:
- Sharing a document publicly: Deep. Removes title, author, and keywords so the file carries no identifying fields.
- Sending to a client or colleague: Standard. Drops the generating-application details while keeping the document's own title.
- Keeping the document's metadata attached: Light. Only rewrites the structure and removes nothing.
- Checking whether a file can be smaller: Any level. The result page shows the real before and after sizes so you can see how much is removable metadata versus embedded images the tool leaves untouched.
Common Mistakes That Waste Time
Even with a good tool, people make the same errors repeatedly:
Expecting Big Reductions on Image-Heavy Files
This tool does not re-encode embedded images. If most of your file's size comes from scans or photos, the reduction will be small — the before and after sizes on the result page make that clear right away.
Ignoring the Source Document
The easiest way to get a small PDF is to create it correctly in the first place. Most applications let you choose image quality and whether to embed full fonts or subsets during export.
Using Screenshots Instead of Native Exports
Taking screenshots of a document and pasting them into a new PDF turns crisp, selectable text into a raster image — which then needs to be high-resolution to remain legible, resulting in a much larger file.
Scanning at Unnecessarily High Resolution
If you're scanning physical documents to PDF, 150 DPI is sufficient for text-based pages. 300 DPI is the standard for pages with photos. Scanning at 600 DPI or higher creates enormous files with no practical benefit for most use cases.
Expecting Quality Changes After Cleanup
Because text and images are copied through unchanged rather than re-encuded, the cleaned file looks identical to the original. There is no quality loss to check for — but do confirm the file opens correctly.
Why This Tool Does Not Subset Fonts or Recompress Images
Two techniques shrink PDFs dramatically — subsetting fonts down to only the glyphs used, and re-encoding embedded images at lower quality. Neither is applied here, and that is a deliberate trade-off worth understanding.
Both techniques change the document's contents. Subsetting replaces the full typeface; recompressing images permanently discards pixel data. This tool instead preserves text and embedded images exactly as they are, which means it never degrades quality — but it also means a file dominated by embedded images or fully embedded fonts will shrink only modestly, since those bytes are left untouched.
For most files the metadata removal and structural rewrite are enough. If you need a much smaller file and are willing to trade away selectable text and image fidelity, you would need a rasterizing converter that re-encodes the whole document — a different tool with a different trade-off.
Beyond Compression: Other Ways to Shrink a PDF
Sometimes compression alone isn't enough. A few alternatives:
- Split the PDF. If you only need to send specific pages, split the PDF and send just the relevant section.
- Convert images separately. If images are the bulk of the file, convert PDF pages to JPG and share them separately, keeping the text PDF lightweight.
- Merge after cleaning. If you're combining multiple PDFs, clean each one's metadata first, then merge them so the final combined file carries no stray author or application fields.
- Use a different format. For purely text documents, consider DOCX or plain text. For purely visual documents, consider image formats.
Conclusion
Reducing a PDF's size without touching its content comes down to one principle: remove what is hidden and leave what is visible alone. Metadata and the file's internal structure carry no visual information, so they can be cleaned up without any quality trade-off.
Most PDFs carry leftover metadata from the tools that generated them — author names, producer strings, internal timestamps. A cleanup tool like Compress PDF removes those fields and rewrites the structure, and the result page tells you exactly how much smaller the file became.
For files whose size comes mainly from embedded images, the honest answer is that metadata cleanup helps only a little — the before and after sizes will show you, rather than a promised percentage. That transparency is more useful than an inflated claim.