You have a document to submit and a form that refuses anything over 5 MB. Or 2. Or 10, on a good day. Your file is 40.
The situation is always the same: the limit comes from outside, you cannot argue with it, and you have to fit under it without ruining the document.
Here is why PDFs get big, what actually makes them smaller, what each method costs you, and how to recognise a file that is already as small as it will go.
The weight is almost always pictures
Ten pages of plain text come to a few hundred kilobytes. The same ten pages, scanned, come to 30 megabytes.
The difference is that in a scan each page is no longer text — it is a photograph of a page. A good-quality photograph at 300 dots per inch takes several megabytes on its own.
The same thing happens with a deck full of images, a proposal with renders, or a submission where you pasted screenshots.
So the first question worth asking is whether your document contains images at all. The answer decides whether there is anything left to compress.
There is an easy check: try to select a word with your mouse. If it highlights, it is text. If nothing selects, you are looking at a picture, and the document is a scan.
What actually makes a PDF smaller
Re-encoding the images. The photographs inside are saved again at lower quality or a smaller resolution. The text stays text: still selectable, searchable and copyable. This is what you want in most cases.
Turning pages into pictures. Every page becomes a single image. The saving is dramatic and the text stops being text — only a drawing of it remains. It cannot be searched, a screen reader cannot read it, and whoever receives the file cannot copy anything out of it.
For a document that was already scanned, the second method is often the right call: the text was never text to begin with. For a contract or a proposal you wrote yourself, it is a real loss.
What does not make a PDF smaller
Zipping it. Put a PDF in a zip archive and almost nothing happens. A PDF is already compressed internally, so there is nothing left to squeeze. You might gain two or three per cent, which never gets you under a limit.
Stripping metadata. The title, author and date take a few bytes. Remove them for privacy, not for size.
Printing to PDF. "Print to PDF" repackages the document. Sometimes it comes out smaller, often larger, and it very often turns the text into an image without warning you. It is the most common way people lose the text in a document without noticing.
There is a floor you cannot go below
If your document is 200 pages of text with no images, you are not getting it to 2 MB. Text, embedded fonts and the document's own structure take what they take. There is nothing to re-encode, because the weight is not photographs.
At that point there are two honest ways out: split the document into several files and send them separately, or accept turning the pages into pictures and lose the text.
The reverse is worth knowing too. If somebody already optimised the file with a proper tool, any other tool will make it bigger, not smaller — rebuilding a PDF throws away the work the previous writer did. If the result comes out heavier than the original, keep the original.
Doing it without uploading the file anywhere
This is the part worth deciding before you start rather than after.
To make a PDF smaller, nearly every site asks you to upload it. Often the document is a contract, a bank statement or a scan of an ID. You never learn who opens it, where it ends up, or how long it stays.
For simple operations — compressing, merging, splitting, reordering — the file does not need to leave your device. All of it can happen in the browser.
We built a free tool that does exactly that. You tell it the size you are allowed — 2, 5, 10 MB, or your own number — and it looks for the best quality that fits underneath. You choose whether to keep the text or accept turning the pages into pictures. The file is never uploaded.
No account, no email address, no size limit.
In short
Check first whether the document contains images. If it does not, compression will not help and you need to split it instead.
If it does, re-encode the images and keep the text. Only turn pages into pictures if the document was a scan anyway, or if you have run out of options.
Do not waste time zipping it, do not print to PDF to shrink it, and if the result is bigger than the original, keep the original.
And where you can, do this work locally. A confidential document has no reason to pass through somebody's server for an operation your own browser can finish on its own.
