Compressing a PDF makes it smaller, but "smaller" covers a range of things happening under the hood. What actually gets changed depends on what the file contains and what kind of compression is applied. Understanding this helps you predict what the output will look like โ and explains why the same compression setting produces dramatically different results on different documents.
A PDF Is Several Types of Data in One File
Before getting into what compression does, it helps to know what's inside a PDF. A typical document contains some combination of: vector graphics (shapes, lines, charts drawn mathematically), raster images (photographs, scanned pages, screenshots), text with font data, document metadata (author, creation date, title), and structural information (page layout, bookmarks, links). Each of these is stored differently and responds to compression differently.
What Happens to Images
Images are where most of a PDF's file size lives, and they're where compression has the most dramatic effect. When you compress a PDF, the tool re-encodes embedded images using a more aggressive compression algorithm โ typically JPEG for color images, JBIG2 or CCITT for black-and-white. This reduces the amount of data stored to represent each image.
JPEG compression is lossy โ it permanently discards some image data to achieve smaller file sizes. At high quality settings (80-90%), the discarded data is in areas of the image where the human eye is least sensitive: fine texture, subtle color gradients, high-frequency detail. The result looks essentially identical. At lower quality settings, the discarded data becomes visible as blockiness, blurring, or color banding โ the classic JPEG artifacts.
Downsampling is a related step that some compression tools apply. It reduces the pixel dimensions of images โ a 300 DPI image might become 150 DPI โ which makes files much smaller but also makes images softer if printed or zoomed. Quality PDF Compression tools either skip downsampling or make it an optional setting.
What Happens to Text
Text in a PDF is stored as characters with positions, styles, and font references โ not as images of letters. This data compresses very efficiently using lossless algorithms like Deflate (the same algorithm used in ZIP files). Lossless compression reduces file size without removing any data: the decompressed output is bit-for-bit identical to the original.
This means text in a compressed PDF is exactly the same as in the original โ same characters, same positions, same font rendering. A compressed document's text is not degraded in any way. If text in a compressed PDF looks different, the issue is font rendering in the viewer, not anything the compression changed.
What Happens to Fonts
Embedded fonts can contribute significantly to file size. A PDF that embeds complete font files for five typefaces might have several megabytes of font data alone. Compression tools can address this through font subsetting โ removing from the embedded font any characters that aren't actually used in the document. A document that only uses the letters A-Z and 0-9 doesn't need the full Unicode character set embedded.
Font subsetting is lossless from the reader's perspective โ the document still renders correctly because all the characters that appear in it are still present. The savings depend on how many characters the original font embedded vs. how many are actually used.
What Gets Removed Entirely
Beyond compressing existing data, compression tools also remove overhead that accumulates inside PDFs over time. Every time a PDF is edited and saved, the previous version of edited elements is often retained in the file structure rather than deleted โ this is how undo history and version recovery work inside PDF editors. A document that's been through many edit-save cycles can carry a lot of dead weight: old versions of objects, duplicate data, deleted content that was never actually removed from the file.
Compression cleans this out. The tool rebuilds the file structure from scratch, keeping only the current state of each element and discarding the accumulated history. This is why freshly compressed PDFs are sometimes dramatically smaller than expected โ the file was carrying significant hidden bloat from previous edits.
What Stays Exactly the Same
Compression doesn't change the document's content, layout, or structure. Page count, page order, text content, vector graphics, bookmarks, hyperlinks, and form fields all remain intact. The compressed PDF opens, displays, and behaves identically to the original for all practical purposes.
What might change โ depending on settings โ is image sharpness at high zoom levels and the file's metadata. Some compression tools strip or update metadata, which is usually fine but worth knowing about if the original creation date or author field matters for your records.