Not usually. Most PDFs contain vector text with embedded fonts. A PDF produced by a scanner is different: it is a container holding one photograph per page, which is why you cannot search it until OCR has been run.

Why do PDFs look the same everywhere?

Because the file embeds the fonts and specifies coordinates for every mark on a fixed-size page. The viewer does not make layout decisions, so there is very little left to differ between devices.

Structurally, yes — pages, annotations, form values and metadata can all be changed. Rewriting a paragraph is harder, because the file stores positioned glyph runs rather than sentences.

What a PDF actually is (and why it refuses to die)

A PDF is a description of where marks go on a fixed-size page, not a document that reflows — which is exactly why it survived thirty years of format churn.

A PDF is often described as "a picture of a document". That is wrong in a way that causes real problems, because it leads people to expect image-like behaviour from a file that is usually nothing of the sort.

The page-description model

At its core, a PDF is a list of drawing instructions bound to fixed-size pages. A typical page says something close to: select this embedded font at 11 points, move to coordinate (72, 708), draw this run of glyphs, move to (72, 694), draw the next run, then place this JPEG in the rectangle from (72, 300) to (523, 600).

Three consequences follow, and they explain most PDF behaviour you will ever encounter.

Layout is frozen. The file does not describe paragraphs that reflow to fit a window; it describes positions. Open the same PDF on a phone, a laptop and a print shop's RIP and the marks land in the same places relative to the page. This is the entire reason the format is used for contracts, invoices, tickets and academic papers.

Fonts travel with the file. A PDF normally embeds a subset of each typeface it uses — only the glyphs the document actually needs. That is why PDFs are self-contained, and also why a document using five weights of two font families is meaningfully larger than one using a single weight.

Text exists as positioned runs, not as sentences. The file knows a glyph sequence sits at a coordinate. It does not necessarily know that two runs belong to the same paragraph, that a block is a table cell, or which of three columns to read first. Extraction tools reconstruct that reading order from geometry, and they get it wrong on complex layouts because the information was never stored.

Where the size goes

People are frequently surprised by a 40 MB PDF that "only has fifteen pages". The text is almost never the cause. In a text-only document, a page of prose is a few kilobytes.

Weight comes from embedded raster images. A single A4 page scanned at 300 dots per inch is roughly 2480 × 3508 pixels — about 8.7 megapixels, comparable to a photograph. Fifteen of those, stored with light compression, easily reach tens of megabytes. The PDF Info tool will show you the page count and metadata, and if the file is large relative to its page count, images are the reason. Our guide on reducing PDF size covers what to do about it.

Two very different kinds of PDF

This distinction matters more than any other in day-to-day work.

A born-digital PDF was exported from software — a word processor, a spreadsheet, a design tool, a reporting engine. It contains real text objects. You can search it, select text, copy a quote, and convert it to Word with reasonable results.

A scanned PDF was produced by a scanner, a phone camera or a fax bridge. Each page is one photograph. Visually it may look crisp and typeset, but there are no characters in the file at all — searching returns nothing and selection tools find nothing to select. Making it searchable requires OCR, which analyses the pixels and writes a recognised text layer behind the image.

Plenty of files are hybrids: a born-digital report with a scanned signature page appended, for example. Checking before you act saves time; see how to tell whether a PDF contains real text.

What else is in the file

Beyond page content, a PDF carries structures that are easy to forget about:

Document metadata — title, author, subject, keywords, producing application, creation and modification timestamps.
Annotations — comments, highlights, stamps and link areas, stored separately from page content.
Form fields — AcroForm or XFA fields with their own values, which are not part of the page drawing instructions.
Attachments — entire files can be embedded inside a PDF.
Incremental update history — because PDFs can be saved by appending changes rather than rewriting, an earlier state of the document can survive in the same file.

That last two points are why PDFs leak information. If you are sending something sensitive, read what is hidden inside a PDF before you attach it.

Why it is still the default

Every few years a format arrives that is better at something PDF is bad at: reflowing on small screens, collaborative editing, semantic structure. None of them replaced it, because the PDF's weakness — its rigidity — is the property institutions actually want. An invoice that renders differently on the recipient's machine is a dispute. A contract whose pagination shifts is a problem. A regulatory filing that reflows is unacceptable.

PDF also has an unglamorous advantage: it is an ISO standard (ISO 32000) with decades of independent implementations, including open-source ones. A file written today will still open in twenty years, which is not a safe assumption for proprietary formats.

Practical takeaways

Before doing anything to a PDF, work out whether it contains real text or photographs of text. Everything downstream depends on it.
If a file is unexpectedly large, look at images, not text.
Do not expect conversion back to an editable format to recover the original document — the structure that a word processor relies on was discarded when the PDF was produced.
Treat the file as carrying more than its visible pages, and clean it before sharing.