A ten-page document should be a few hundred kilobytes. When it arrives as 40 MB, something specific is responsible — and it is almost never spread evenly across the file. Finding the culprit takes about a minute and tells you which fix will work, instead of running compression three times and wondering why nothing changed.
The short version
Work through these in order and stop when one matches:
- Is it a scan or made from photos? Then it is images, and the fix is resolution, not compression.
- Does it have far more pages than you need? Then the fix is removing pages or splitting.
- Was it exported from a design or office application? Then it is probably embedded fonts and vector artwork.
- Is it a form, or does it have revision history? Then it may be carrying things you cannot see.
How to tell which one it is, in under a minute
Divide the size by the page count. This single number identifies the problem most of the time.
| Per page | Almost certainly | What will work |
|---|---|---|
| Under 100 KB | Text and vectors | Nothing to fix |
| 100–500 KB | Normal document with some images | Standard compression |
| 1–5 MB | A colour scan, or photos at full resolution | Lower the image resolution |
| Over 5 MB | Full-resolution camera photos, one per page | Resize the images, or rebuild the PDF |
Then scroll through it and watch what you see. If you can select the text with your cursor, it is a real text document. If you cannot — if the whole page behaves like one picture — it is a scan, and every page is an image file wearing a document costume.
Cause 1: scanned pages, at the wrong resolution
This is the overwhelming majority of oversized PDFs.
Scanners default high — 300 DPI in full colour is common, and some default to 600. At 300 DPI a single A4 page is roughly 8.7 million pixels. In colour, that is a photograph per page, and ten of them is a 30 MB file behaving exactly as designed.
What resolution do you actually need?
- 150 DPI is comfortable for reading on screen and prints acceptably on an office printer.
- 300 DPI is worth it for a document going to a professional print process, or where fine print genuinely has to survive.
- 600 DPI is for archival reproduction and almost never for something you are emailing.
And colour is the expensive part. A greyscale scan of a black-and-white document is typically a third the size of the colour version and looks identical, because there was no colour in the original to lose.
If a scan is still too large after that, the real fix is upstream: scan it again in greyscale at 150 DPI. Compression can only throw away detail that was captured; scanning sensibly means never capturing it.
Cause 2: pages you do not need
Worth checking before you compress anything, because it is free.
Bank statements, contracts and downloaded reports routinely carry pages of terms you were not asked for. If an upload wants three pages of a forty-page statement, removing thirty-seven of them will beat any compression setting ever invented.
The same applies to blank pages from a duplex scan, cover sheets, and the “this page intentionally left blank” filler that still costs bytes.
For a hard email limit, splitting is often better than compressing at all — three 8 MB parts get through a 10 MB cap without touching a single pixel.
Cause 3: embedded fonts
Less common but genuinely surprising when it hits.
A PDF has to look the same on a machine that does not have your fonts, so it carries them. Subsetting — embedding only the characters actually used — keeps this small, and most exporters do it. Some do not, and a document using several weights of a couple of families can pick up multiple megabytes of fonts for a two-page memo.
The tell is a PDF that is large, has no images, and has few pages. If you can re-export it, look for a “subset embedded fonts” option in the export settings.
Cause 4: things you cannot see
A few less obvious passengers:
- Invisible OCR text. Scanning software often adds a searchable text layer behind the image. This is genuinely useful and worth keeping — but it is not free, and if you never search these documents it is pure weight.
- Form fields and scripts. Interactive PDFs carry machinery beyond the visible page.
- Revision data. Some editors save changes by appending to the file rather than rewriting it, so an edited document can carry several older versions of itself.
- Attachments. A PDF can have other files embedded inside it.
What each fix is actually worth
| Fix | Typical reduction | Costs you |
|---|---|---|
| Remove unneeded pages | Proportional to pages removed | Nothing |
| Greyscale instead of colour | 50–70% on scans | Colour, if there was any |
| 300 → 150 DPI | 60–75% on scans | Detail at high zoom |
| Standard compression | 20–50% on mixed documents | Little to nothing |
| Re-scan properly | 70–90% | Time, and access to the paper |
| Split into parts | Nothing overall | Recipients get several files |
The pattern is worth noticing: the biggest wins come from not storing something, not from squeezing it. Compression is the last step, not the first.
When to give up and rebuild
Occasionally a PDF resists everything — usually one assembled by an old system or a fax gateway, storing pages in an inefficient format with no metadata to work with.
The blunt approach is to export each page as an image, then build a fresh PDF from those images at a sane resolution. You lose selectable text and any interactivity, and it is genuinely a last resort. But for a document you only need to look at, it reliably produces a small, clean file where nothing else would.
Before you send it
- Open the result and check page one and the last page, not just the thumbnail.
- If it is a scan, confirm the smallest print is still readable.
- Confirm the size is genuinely under the limit rather than close to it.
- If it is going to a portal rather than an inbox, check the accepted formats — some reject PDFs above a page count as well as a size.