Two documents wearing the same file extension

PDF describes two things that have almost nothing in common, and the file extension does not distinguish them. Nearly every question that begins "why can't I just…" is really a question about which one you are holding.

The first kind

A PDF exported from a word processor, a spreadsheet, a design tool or a browser is a set of instructions. Draw this glyph here, in this font, at this size. Fill this rectangle. Place this vector path.

It is small, it scales to any size without blurring, its text can be selected, searched, copied and read aloud, and every word in it is a word.

The second kind

A PDF from a scanner, a phone camera or a fax is a stack of photographs. Each page is one image, wrapped in the minimum PDF needed to hold it.

It is large, it blurs when enlarged, and it contains no words at all. Every letter you can see is a pattern of light and dark pixels that your eye resolves into text and the file has no idea about.

Why this decides everything

• Searching. The first kind is searchable. The second is not, until it has been through OCR, and even then you are searching a guess.
• Compressing. The first kind is already tiny; compressing it means turning it into the second kind, which makes it larger. The second kind compresses well, because there is image data to throw away.
• Editing. Adding a watermark or a page number works on both. Changing a word in the text works on neither, but for different reasons.
• Accessibility. A screen reader can read the first kind. It has nothing at all to work with on the second.
• File size. A forty-page report might be 300 KB. The same forty pages scanned might be 40 MB. Both are "a PDF".

How to tell in five seconds

Open the document and try to select a line of text with your cursor.

If a neat highlight follows the words, it is the first kind. If you get a rectangular selection box that ignores where the letters are — or nothing happens at all — it is the second.

That single test predicts more about what you can do with a file than anything else you could check.

The awkward middle

Plenty of documents are both. A scanned contract with a digitally added cover page. A form filled in on screen, printed, signed and scanned back. A report where someone pasted in a screenshot of a table.

These are the files that produce confusing results, because a tool that examines the first page draws a conclusion about the whole document. An extraction that returns three pages of text and then nothing usually means the document changed kind partway through, not that the tool failed.

Inspecting a document before working on it — how many pages have a text layer, how many sizes are in use, whether it has forms or bookmarks — takes seconds and tells you which tools are even applicable. It is the least glamorous step in any document workflow and the one that saves the most time.