Why did compressing my PDF barely change the size?

Because it is a text PDF. Vector text and embedded fonts are already compact and already compressed internally. There is very little to remove — the size is structural, not image data. Removing pages or splitting the document is the only real lever.

Does compressing a PDF lose the ability to search it?

It depends on the method. Optimising the images inside a PDF keeps the text layer intact. Compression that rasterises each page — turning the whole page into a picture — destroys selectable text permanently. If searchability matters, check that you can still select text in the result.

What size should I aim for?

Whatever the destination requires. Email attachments are commonly capped around 10–25 MB, and upload portals are often stricter. There is no benefit in going smaller than the limit at the cost of legibility.

How to reduce PDF file size without ruining it

File size is almost always images. Find out whether your PDF is text or scan first, because compression helps enormously with one and does nothing for the other.

Almost every oversized PDF is oversized for the same reason, and it is not the one people expect. It is rarely the page count and almost never the text. It is images.

Find out what you actually have

Before compressing anything, work out which kind of PDF you are holding, because the answer determines whether compression will help at all.

Open the file and try to select a line of text with your cursor. If individual words highlight, the pages are real text. If the whole page highlights as one block, or nothing does, the pages are pictures — a scan. There is more detail on this test in does my PDF have real text.

Then divide the file size by the page count. A text PDF typically sits between 30 KB and 300 KB per page. Anything above roughly 1 MB per page means images dominate. PDF Info will give you the page count without opening the file in a reader.

If it is a text PDF

Compression will disappoint you. Vector text is already stored efficiently and the content streams are already compressed with Flate — running them through a compressor again recovers almost nothing. A 4 MB, 300-page report is not a problem to be solved; that is simply what the document weighs.

The levers that actually work here are structural rather than lossy:

Remove what nobody needs. Appendices, duplicate cover pages, blank separator sheets from a scanner. Remove Pages deletes them without touching the rest of the file.

Send only the relevant part. If the recipient needs chapter three, Extract Pages gives them chapter three. This is usually both smaller and more considerate than sending everything.

Split into parts. For a document that genuinely has to travel in full, Split PDF produces several files that each clear the limit.

Check for embedded attachments. Some PDFs carry whole spreadsheets or datasets inside them. These are invisible in normal reading and can account for most of the file.

If it is a scan

Here compression works, and it often works dramatically — reductions of 70–90% are ordinary, because scanners are routinely generous with resolution and colour.

Two things drive the size of a scanned page:

Resolution. A page scanned at 600 DPI holds four times the pixel data of the same page at 300 DPI. For ordinary printed text, 300 DPI is the point past which extra resolution stops being visible on screen or on paper. Anything above that is weight without benefit — unless the document has genuinely small print you may need to enlarge.

Colour depth. A colour scan of a black-and-white page carries three channels where one would do, roughly tripling the size to represent nothing. Greyscale is the sensible default for text documents; keep colour only where colour carries meaning.

Compress PDF works on both of these — it re-encodes the page images at a lower resolution and quality, which is exactly the right operation for a scan and the wrong one for a text document.

The trade-off nobody explains

Compression that rasterises pages does something worth understanding before you accept the result: it replaces the page with a picture of the page.

The visual result can look nearly identical while the document has quietly lost properties you may depend on. Selectable text becomes unselectable. Search stops finding anything. Copy-paste returns nothing. Links and form fields stop working. Accessibility tools have nothing left to read aloud.

If your document was searchable and you need it to stay searchable, verify after compressing that you can still select a word. If you cannot, and it matters, run OCR PDF on the compressed file to add a fresh text layer — a lower-quality one than the original had, but better than none.

A sensible order

Establish whether it is text or scan. This decides everything that follows.
Remove pages that do not need to be there. Free, lossless, and often sufficient on its own.
Compress only if images dominate. Start with a moderate setting rather than the most aggressive one.
Open the result and read the pages that matter at 100% zoom. Compression artefacts show up first on small text inside images and on thin lines in diagrams — precisely the content people need to read.
Confirm the text is still selectable if it was before.
If it is still too large, split it rather than compressing harder. A second aggressive pass usually costs far more legibility than it saves bytes.

When to stop

Compression has a floor. Past a certain point you are trading readability for a number, and an illegible 2 MB document serves nobody better than a legible 12 MB one. If a document will not fit under the limit while remaining readable, the answer is to send fewer pages, split the file, or share it another way — not to compress it into uselessness.

For the specific case of getting a file under an email or portal cap, preparing a document for email walks through the whole sequence.