What actually makes a PDF large
A PDF is a container, and what is inside determines everything about its size. There are two very different kinds of document, and only one of them can be meaningfully compressed.
A text PDF, exported from Word, Pages or a web page, stores letters as instructions: "draw this character, in this font, here". A hundred pages of that might be 200 KB. There is almost nothing to compress, because there is almost nothing there.
A scanned PDF stores a photograph of each page. A single page scanned at 600 DPI is around 25 million pixels. This is where the megabytes are, and it is what this tool targets.
So if your PDF is 12 MB, it is a scan, and compressing it means reducing the resolution and quality of those page images until the total fits your limit.
Aiming just under the limit
Most PDF compressors offer "low, medium, high" and leave you to guess. Guessing badly in one direction leaves you still over the limit; guessing badly in the other gives you an unreadable document when you had room to spare.
This tool searches instead. It steps down the quality and resolution of the images (the embedded pictures in a text document, or the page images of a scan), rebuilding the document and measuring the real byte count each time, and keeps the largest result that still fits. If you ask for 1 MB, you should get something close to 1 MB, not 180 KB of illegible grey.
The trade-off you are accepting
Re-rendering pages as images happens for a scan, and for a text document only if you turn off "Keep my text selectable". When it does, it is destructive in ways that are not always obvious:
- Text is no longer text. It cannot be selected, searched, or read aloud by a screen reader. If accessibility matters, this is a real cost.
- Links and form fields are gone. Everything becomes a flat picture.
- Very small text degrades first. Footnotes and fine print are the first casualties.
Check the result before you rely on it. If the original was already a scan, you have lost nothing that was not already lost.
Better: scan properly in the first place
If you control the scan, these settings avoid the problem entirely:
| Setting | Use | Why |
|---|---|---|
| Resolution | 150–200 DPI | Ample for text. 600 DPI is nine times the data for no visible gain. |
| Colour mode | Greyscale, or black & white for plain text | Colour triples the data for a document that is black ink on white paper. |
| Format | PDF, not TIFF | TIFF is often uncompressed and enormous. |
Phone scanning apps default to high-resolution colour, which is why a three-page certificate arrives as 15 MB.
When compression is the wrong answer
If your document is 40 pages and the limit is 500 KB, no amount of compression will produce something readable. The portal is telling you it expects a single certificate, not a bundle.
Use the split tool to pull out the pages actually required. If the issue is the opposite, several separate files where one is wanted, the merge tool combines them, and images to PDF turns photos of documents into a single file.