Compress a PDF to under 1 MB

A megabyte is a lot of budget

For most documents this is the target where compression stops being a trade-off. A ten-page scan gets around 100 KB a page, which is enough for 150 DPI at a JPEG quality in the seventies — the point at which the difference from the original is invisible on screen and barely detectable in print.

The interesting question at 1 MB is usually not "will it look bad" but "why was my file 47 MB in the first place".

Where the original size came from

PDFs are almost never big because of text. A forty-page report exported from a word processor is often under 300 KB in total, because each page is a few kilobytes of drawing operators plus a shared font. Size comes from images, and specifically from these:

• Phone-camera scans. A modern sensor produces a 12-megapixel JPEG per page. At three or four megabytes each, twelve pages is 40 MB before anything else.
• Flatbed scans at 600 DPI. Four times the pixels of 300 DPI for no benefit unless you are digitising artwork.
• Losslessly compressed images. Some scanners emit PNG or Flate-compressed bitmaps, which are enormous compared with a JPEG of visually identical quality.
• Embedded fonts, repeated. Merging twenty documents can embed twenty copies of the same font subset. This is small in comparison but it adds up.

Re-rendering at 150 DPI removes roughly 90 per cent of the pixels from a 600 DPI scan before the JPEG encoder does any work at all. That is why the reduction figure is often 95 per cent or more without any visible loss.

1 MB targets you will actually meet

• Web forms built on a CMS, where 1 MB is a common default upload limit.
• Learning-management systems collecting coursework.
• Court e-filing portals, several of which specify 1 MB per exhibit.
• Older intranet document libraries and SharePoint list attachments.
• Anything sent over a slow or metered connection.

Checking the result properly

The per-page table under the result is worth two seconds of attention on a document this size. What you are looking for:

1. Any page with a scale below 100%. That page could not fit even at minimum quality and was reduced in resolution. On a 1 MB target this usually means the page is a full-bleed photograph.
2. A big spread of quality values. That is the budget roll-over working — a title page finishing at 94 and a dense chart page at 55 is the algorithm doing exactly what it should.
3. Everything at 94. The maximum. Your document had headroom to spare, so you could have asked for a smaller target, or you did not need to compress at all.

Very large source files

There is no upload, so the input size limit is your device's memory rather than a plan tier. In practice:

Only one page is held in memory at a time, so page count matters far less than the size of any individual page's image data.