Most PDF compressors offer one button and one promise: smaller. That simplicity hides a choice that matters, because "smaller" can be achieved in two fundamentally different ways, and they do not preserve the same things. One way tidies the file's structure and removes redundant data while keeping every page exactly as it is, text included. The other rebuilds each page as an image, which can shrink much further but turns text into pictures of text. A tool that shows both options honestly, with real byte counts before and after, lets you choose based on what you need the file for afterwards.

Compress PDF does exactly that. Its lossless mode re-saves the file without rebuilding pages, so text stays selectable and searchable. Its smallest-size mode rebuilds pages as images, trading text for size. The rest of this post runs one fictional document through both modes, so you can see the difference in numbers and in consequences, rather than taking the distinction on trust.

What lossless compression actually does

A PDF accumulates clutter. Repeated edits, exports, and saves leave behind redundant objects, unreferenced data, and inefficiently stored structures. A lossless re-save rewrites the file cleanly: the same pages, the same images, the same text, stored without the accumulated waste. Nothing visible changes. Text stays selectable, links stay live, and the document still behaves like a document rather than a set of pictures.

The saving from a lossless pass depends entirely on how much clutter the file carried. A freshly exported, well-formed PDF may shrink only slightly, because there is little waste to remove. A file that has been edited, merged, and re-exported several times may shrink noticeably. Neither result means the tool failed or succeeded in any absolute sense. Lossless compression removes waste; it cannot remove content. If the file's bulk is in its images or fonts rather than in waste, the lossless saving will be modest, and that modest result is telling you where the bulk really sits.

What the smallest-size rebuild does differently

The smallest-size mode takes a different approach: it renders each page as an image and builds a new PDF from those images. Because photographs and complex layouts often compress well as images, this can produce a dramatically smaller file, especially for image-heavy or scanned documents. The cost is equally clear. Text in the rebuilt file is no longer text. You cannot select it, search it, copy it, or have a screen reader read it as text. The document looks the same at a glance, but it has changed in kind: from a structured document to a picture book of its own pages.

That change has practical consequences beyond searching. A rebuilt file cannot be edited as text later, its contents cannot be indexed by a document search, and any accessibility that depended on the text layer is gone. For a file that will only be read and printed, none of that may matter. For a file that will be archived, searched, quoted, or read aloud, it matters a great deal. The smallest mode is a powerful option, and it should be chosen deliberately, not by default.

Worked example: one document, both modes, real-style counts

The document in this example is fictional: an invented newsletter, invented figures, and illustrative byte counts chosen to show the pattern, not measured from any real file. Imagine the fictional Willow Lane Gardening Club, contact via hello@example.org, produces a 16-page seasonal newsletter with photographs, member listings, and a calendar of events. The exported file is 22 MB, too large for the club's mailing list attachment limit.

Running the lossless re-save brings the file to about 18 MB. The pages look identical, the text is still selectable, and the member names can still be searched. The saving, roughly a fifth, comes from structural waste, and it is not enough for the mailing limit. Running the smallest-size rebuild on a copy brings the file to about 4 MB, comfortably under the limit. The newsletter reads fine on screen and in print, but searching for a member's name now finds nothing, because the names are pixels, not characters.

Faced with those two results, the club makes a split decision, which is the sensible pattern. The 4 MB rebuilt version goes out to the mailing list, where the goal is simply that everyone can open and read it. The 18 MB lossless version, and the 22 MB original behind it, stay in the club's archive, where searchability matters for future years. Neither version is "the" compressed file. Each serves the purpose it fits, and the byte counts shown by the tool make the trade-off visible instead of hidden.

How to choose between the modes

Ask one question first: will anyone need to search, copy, or select text in this file after compression? If the answer is yes, or even maybe, start with the lossless mode and accept whatever saving it gives. If that saving is not enough, look for other savings that preserve text: removing unneeded pages, or optimising images before they enter the PDF, rather than rebuilding the whole document as images.

If the answer is clearly no, because the file will only be read or printed, the smallest mode is a legitimate choice, and the size saving is often dramatic. Even then, work on a copy and keep the original. Compression decisions are easy to reverse when the original survives, and impossible to reverse when it does not. Check the rebuilt result page by page at normal reading size before sending it: photographs should stay clear, and small print should stay legible. If small print turns to mush, the document was a poor candidate for rebuilding, and no size saving justifies an unreadable result.

Honest limitations

Neither mode can shrink a file below the size its content genuinely requires, and neither can improve the quality of what is already there. Lossless mode preserves everything and therefore shrinks least. Smallest mode shrinks most and sacrifices text. A file that is mostly print-quality photographs may still be large after either mode, because those photographs are the content, not waste. In that case, the honest options are accepting the size, splitting the document, or sharing it by link rather than attachment. If your file is scanned and has no text to begin with, OCR PDF can add a searchable text layer first, but be aware that a smallest-size rebuild afterwards would remove that layer again, so order your steps with the end use in mind.

FAQs

What is the difference between lossless and lossy PDF compression?

Lossless compression, the re-save mode, removes structural waste while keeping every page, image, and character of text intact. Lossy compression in the smallest-size sense rebuilds pages as images, which shrinks further but turns text into pictures that cannot be selected or searched. The words describe what is preserved, not just how much smaller the file gets.

Will compressing a PDF make it blurry?

The lossless mode does not touch image quality, so nothing gets blurrier. The smallest-size rebuild re-renders pages as images, so photographs may soften slightly and very small print may lose sharpness. Always inspect the rebuilt result at normal reading size before sending it.

Will compressing a PDF remove its text layer?

Only the smallest-size rebuild does that, because its pages are images. The lossless re-save keeps text selectable and searchable. If text matters to you, use the lossless mode and keep the original as well.

Which mode should I use for emailing a PDF?

If the file will only be read, the smallest mode usually gets under attachment limits most reliably; check legibility first. If recipients need to search or copy from it, use the lossless mode, and consider the guide to image formats for background on why images dominate file size. PDF Toolkit can also help prepare the file first, for example by rotating or reordering pages before compression.

Related reading: When the file is still too large to email.

human review pending client designation