PDF Still Too Big After Compressing? What Actually Works
Updated · PixKB
You compressed the PDF, the portal still says "file size must be less than 200 KB", and trying again gives the same result. We ran the common failure cases through our own PDF compressor and measured what happened. Here is why a PDF refuses to shrink, and what to do in each case.
Quick fix first: try Compress PDF to 200 KB (or any limit). It aims at your exact number, so if your previous tool used a fixed "medium" setting, this alone may be enough.
1. Too many pages for the limit
A size limit is shared by every page. Two scanned pages in 200 KB means about 100 KB each, which is plenty. Ten pages in 200 KB means 20 KB each, and that's too little for a readable scan.
We built a 10-page phone scan (34 MB) and compressed it:
| Limit | Result | Page images kept at |
|---|---|---|
| 200 KB | not reachable readably; smallest 276 KB | 600 px |
| 500 KB | 477 KB | 751 px |
| 1 MB | 996 KB | 1091 px |
What works: upload only the pages the form asks for (blank back sides and instruction pages waste budget), ask whether the limit is per file or per page, or split the document across upload boxes if the portal has several. As a rough guide, keep at least 40–50 KB per scanned page for normal-size print to stay readable.
2. The PDF is text, not pictures
Compressors shrink images. A PDF exported from Word or a form generator is mostly text and fonts, which are already efficient. Our 30-page text-only test file was 27.4 KB; the smallest it could get with its text intact was 20.3 KB, so a 20 KB limit was out of reach.
Turning such pages into pictures makes things worse, not better: our forecast for the same 30 pages as page images was about 1.35 MB, roughly 65 times bigger. That's why PixKB doesn't even offer that option for text-only files. What works: remove pages, or export again from the original program with fewer pages, smaller embedded images, or "standard" fonts instead of embedded decorative ones.
3. Images stored in print colours (CMYK)
PDFs made for printing sometimes store photos in CMYK, the four-colour format printers use. Recompressing CMYK images safely in a browser is unreliable (colours can invert or shift), so PixKB leaves them untouched and tells you so. In a test file with one CMYK image, the smallest version with text intact was 1.33 MB. With the user's permission, PixKB rebuilt the pages as images and reached 564 KB, under a 700 KB limit.
What works: accept the page-image rebuild (the pages look the same, but text stops being selectable), or export the PDF again from its source program in RGB / "screen" quality.
4. KB isn't always the same KB
Some portals count 1 KB as 1,000 bytes and others as 1,024. A file of 20,300 bytes is "19.8 KB" to one system and "20.3 KB" to another, so it passes one 20 KB limit and fails the next. PixKB always counts the strict way (1 KB = 1,000 bytes), so a file it calls "under 200 KB" passes either rule. If you used another tool and you're within a few KB of the limit, that's probably your problem: aim 5–10% under.
To see the exact size: on Windows, right-click the file → Properties (look at the number in bytes); on a Mac, select it and press Cmd + I.
5. Compressing the already-compressed file again
People often run a file through a compressor two or three times. With a tool that targets a number, that isn't necessary, and with fixed-setting tools each pass loses detail. In our test, the original 6.3 MB scan compressed straight to 100 KB gave 96.5 KB with 751-pixel page images; compressing to 300 KB first and then to 100 KB gave 93.5 KB at the same resolution. Similar result, but every JPEG pass discards a little more detail, so always start from the original file when you still have it.
6. Greyscale won't rescue it
"Scan in black and white" is common advice. We measured it. On a colourful certificate (tinted background, coloured seal), greyscale gave about 8% more resolution at the same 100 KB. On normal black-on-white documents it made no measurable difference after compression. It's a small help, not a fix for a file that's several times too big.
A checklist that works
- Start from the original scan or export, not an already-compressed copy.
- Remove pages the portal didn't ask for. Use merge PDF with a page range like
1-2to keep just those. - Compress to the portal's exact number with a tool that targets it.
- If text can't fit, decide whether page images are acceptable for this upload (fine for most certificates, bad for forms that must stay fillable).
- Check the result's size in bytes, and stay 5–10% under if the portal is strict.
Test notes: all files processed in Chromium on a laptop with PixKB's compressor; sizes in decimal units (1 KB = 1,000 bytes). Test documents were generated for testing, not real personal documents.
Frequently asked questions
Why does my PDF stay the same size after compressing?
Usually because it's mostly text and fonts rather than images, so there's little for a compressor to remove. Removing pages is then the only real way to make it smaller.
How many pages can fit in 200 KB?
For readable scans, about four to five pages at most (40–50 KB per page). Two or three pages fit comfortably. Ten pages did not fit readably in our test; the smallest was 276 KB.
Will converting the PDF to images make it smaller?
Only if the PDF contains large photos or print-colour (CMYK) images. For text PDFs it makes the file much bigger; in our test, 30 text pages became about 1.35 MB as images.
The portal says my file is too big, but it shows under the limit on my computer. Why?
The portal probably counts 1 KB as 1,000 bytes while your computer uses 1,024 (or the reverse). Aim 5–10% under the limit, or use a tool that counts strictly.
Is it OK to compress a PDF several times?
It works, but each pass loses some detail. Start from the original file and compress once, straight to the exact limit.