Why is your PDF 40MB when it looks like plain text?

There are two completely different things called a PDF

A PDF that was made by a computer stores your words as text and your typefaces as instructions. Twelve pages of that is typically 60KB to 400KB. It is one of the most efficient things you can send anyone.

A PDF that came out of a scanner, a phone scanning app, or a photocopier stores a full-colour photograph of each sheet of paper. The words in it are not words. They are dark pixels arranged in the shape of words, and the file has no idea what they say.

Those two files look identical on screen and share a file extension, but one is a document and the other is a photo album with a paper theme. Practically every 40MB PDF anyone has ever been confused by is the second kind.

The five-second test

Open the file and try to drag-select a word in the middle of a paragraph, the way you would in a browser. Real text highlights letter by letter. A scan gives you either nothing at all or a single rectangle covering the whole page image.

A second tell: press Ctrl+F or Cmd+F and search for a word you can plainly see. Zero results on a page that obviously contains the word means the file has no text in it, only a picture of text.

A third, if you are still unsure: zoom to 400 percent. Real text stays razor sharp at any magnification because it is being redrawn, not enlarged. A scan turns into soft grey mush, and you may see faint speckles from the paper and the scanner sensor.

This matters because the fix is completely different for each. You cannot compress your way out of a scan, and you do not need to compress a text file at all.

Culprit one: scanned pages, usually at a silly resolution

Office scanners and phone apps love to default to 600 DPI in full colour. At that setting an A4 page is roughly 35 million pixels, which is a bigger photograph than most cameras produce. Multiply by 60 pages and 40MB stops being surprising and starts being inevitable.

For ordinary typed documents, 600 DPI buys you nothing a human will ever see. 200 DPI is entirely readable, 300 DPI is generous, and both are considered fine for archival and for text recognition. Dropping from 600 to 200 DPI cuts the pixel count by about nine times.

The other free win is colour. A black-ink page scanned in full colour carries three channels of information to describe something that is, in reality, black and white. Rescanning in greyscale typically saves 50 to 70 percent, and scanning as pure black and white can be far more dramatic again, though it is unkind to photographs and pencil marks.

If rescanning is not possible because the paper is gone, you can still rebuild the file: pull the pages out as images with PDF to JPG, shrink them with the image compressor, then reassemble them using images to PDF. It is a slightly clumsy round trip and it is often the single biggest saving available.

Culprit two: images that are far bigger than the space they occupy

The classic version of this is a report where someone dropped a 12-megapixel phone photo into a document and dragged the corner until it was the size of a postage stamp. The page shows a stamp. The file still contains all 12 megapixels.

PDFs almost never resample images on your behalf. Whatever was placed in the source document is what gets embedded, so five product photos straight off a camera can add 30MB to a file that is otherwise a page of text.

The cure lives upstream. Resize images to the dimensions they will actually be printed or viewed at before you place them, then export. A photo shown at 10cm wide needs about 1200 pixels across for print quality and around 600 for screen. Anything beyond that is weight you are carrying for no one.

A useful sanity check when a number sounds abstract: run the file size through a data storage converter and divide by the page count. If a text-looking document is costing more than about 300KB per page, it contains pictures.

Culprit three and four: fonts and everything you already deleted

Embedded fonts are real but usually minor. A full typeface family adds a few hundred kilobytes, and a document set in several weights of several families might carry 1MB to 3MB. That is worth knowing about and almost never the reason a file is 40MB.

Revision history is the sneaky one. Many editors save changes by appending to the end of the file rather than rewriting it, so a document you have edited thirty times can still contain the images you removed, the pages you deleted, and the old version of every paragraph you rewrote.

This is why a file sometimes grows after you delete things from it. It is also why exporting a fresh copy, or running the pages through a tool that rebuilds the document rather than patching it, can drop the size by half with no visible change at all.

The 25MB wall, and the better way round it

Gmail and Outlook both cap attachments at 25MB, and encoding overhead means the real ceiling is closer to 24MB. Plenty of corporate mail servers are stricter, at 10MB or even 5MB, and they tend to reject silently.

The reflex is to compress until the file squeezes under the limit. For a scan, that is how you turn a legible document into grey soup, and the recipient has to ask you for it again anyway.

The better move is almost always to send less. Use PDF split to break a 200-page scan into four fifty-page parts that each sail through, or remove pages so you send the eight pages someone asked about rather than the whole binder.

If you do need to reorganise before sending, reorder pages and merge handle the assembly, and PDF to text will pull the words out of a text PDF when the recipient only wants the content and not the layout.

There is a floor, and pretending otherwise wastes your afternoon

A 200-page colour scan will not become a readable 2MB file. That is 10KB per page for a full photograph of a sheet of paper, and no setting achieves it without destroying the thing you were trying to send.

Realistic targets: a text PDF should be 20KB to 60KB per page. A greyscale scan at 200 DPI lands around 50KB to 150KB per page. A colour scan at 300 DPI is more like 300KB to 800KB per page. If you are already near those numbers, the file is not bloated and further compression only costs you legibility.

So diagnose before you squeeze. Find out whether you have text or photographs, check whether the resolution was ever justified, and consider whether the recipient needs all 200 pages. Those three questions solve the problem far more often than any slider labelled quality.