Why your PDF is large, and what will actually make it smaller
A document that looks like a few pages of text can arrive as a 40 MB file, and nothing on screen explains it. PDF files are rarely large evenly: one thing usually accounts for nearly all of the size, and which one it is decides whether compressing helps.
What is actually taking up the space
Pictures carrying more detail than the page shows
This is the usual answer. A photo saved 3,000 pixels wide and placed six inches wide on the page carries about 500 dots per inch, where 150 is enough for everyday printing. The extra pixels are stored in full and never seen.
Scanned pages, which are pictures of text
A scan holds no text. Each page is one large picture, so the size is the scanner's resolution multiplied by the page count. The format matters more than the resolution: color and grayscale scans are normally JPEG and can be written again smaller, while black-and-white scans use formats Compress leaves untouched.
Embedded fonts
Fonts are embedded so the document looks the same on every machine. A file using several families across several weights, each embedded whole rather than only the characters used, can be mostly fonts. PDFCheetah tells you when most of your file is fonts rather than pictures, and that shrinking those is not available yet.
Leftovers from earlier saves, and text stored uncompressed
Some apps record a change by adding to the end of the file rather than rewriting it, so older copies of what you edited stay behind. The block describing the document is often stored as uncompressed text: in one 152-page file we measured, it was 12.1% of it.
What shrinks each one, and what does not
Nearly all of the saving comes from one place: writing the pictures again at the size the page uses. Everything else is small beside it, and some of it is nothing at all.
JPEG pictures — the whole game
Compress works out how wide each picture is really drawn and writes it again at the resolution that width needs. One already at or below it keeps every pixel and is only re-saved at the chosen quality.
Pictures stored any other way — untouched
Only JPEG pictures can be re-saved. Images in PNG-style compression, LZW, JPEG 2000, fax or two-tone formats pass through byte for byte, as do pictures with transparency or spot inks. A four-color picture that can be re-saved comes out in screen color, with a warning that it may print differently.
Embedded fonts — nothing
Fonts are measured and reported to you, but never reduced. If your document is large because of its fonts, nothing here changes that today.
Leftovers and storage — a little, and nothing deleted
Compress removes nothing for being unused. It packs the file more tightly on the way out and compresses that uncompressed description block, and rewriting leaves one current copy of each page rather than a stack. On a text-heavy document that is the only lever, and often not enough to be worth offering.
PDFCheetah will not offer an option that saves less than 5% of your file. If nothing clears that bar it says so, and names the reason for each picture it left alone, before anything runs.
What to expect from compressing your file
Compress measures the file on your device first: every picture, how wide each is drawn, the fonts, whether the document is signed. What it can offer depends on the document.
- A color or grayscale scan
- The best case. Pages scanned at 300 dots per inch as JPEG can be written again at 150 or 100. Check the preview before accepting the smallest option: nothing tests whether small print stays readable.
- A text-heavy report
- Often nothing. With no pictures to write again, the only saving is in how the file is stored, and that frequently falls under the 5% floor.
- A print brochure full of photos
- It depends how it was made. Ordinary JPEG photos in screen color are ideal; print-shop files often are not, since spot inks and transparency are never touched.
- A filled or fillable form
- Usually little: a form is mostly text and line art. Fields, links and annotations survive every option except Flatten to images, which takes the fields with them.
- A password-protected file
- Nothing, by design. Compress will not open a document that needs a password. Remove it in the app that set it, then try again.
Four options at most, often fewer. Best quality touches no picture. Recommended writes pictures again for 150 dots per inch, fine for everyday printing; Smallest file goes to 100 and looks soft in print. Flatten to images turns pages into pictures — the only thing that shrinks a two-tone scan, and the one that costs you selectable text, links and form fields. It is never preselected.
Every size you see is estimated from your own file on your own device, and set slightly high so the finished file comes in under it. A result reaches you only if it is smaller; otherwise you get your original back.
Where to go next
If your document is large because of its pictures, compressing it is the direct answer. If it is large because it is long — a scanned book, a merged bundle — keeping only the pages you need is. Both run in your browser, and neither uploads your file.