A PDF can look perfect on screen and still be unreadable to anything that is not a human eye. Whether it converts cleanly, reads aloud correctly or gets understood by software depends on a layer you cannot see, and nothing on the page tells you whether that layer is there or whether it is any good. Here are three tests that answer the question in about three minutes, with no special software.
What "structured" actually means
Every PDF carries what you see: text placed at coordinates, images, lines. Some PDFs carry a second layer alongside it, describing what those things are. This is a heading. This is a list. This is a table, and these are its cells. This is the order in which the document reads.
That second layer is optional, invisible, and decisive. It is what lets a document reflow on a phone instead of forcing you to pinch and scroll. It is what a screen reader follows. It is what a search index uses to work out which heading a passage sits under. And it is what converters rely on when they rebuild your document in another format.
When it is missing, software has to guess from visual position alone, and it guesses badly. We looked at that failure in detail in why AI tools misread your PDFs. What follows is the practical companion: how to find out where your own files stand.
Test 1: copy a table and paste it
Open the document, select a table or a passage laid out in two columns, copy it, and paste it into a plain text editor or an email draft.
If what lands is coherent, row by row or column by column, the document carries real reading order. If the columns run into each other, if cells arrive in an order nobody would choose, if sentences from two columns get spliced into one line, then there is no structure underneath. What you just saw is what every automated reader sees.
This test takes fifteen seconds and it is the one that tells you most.
Test 2: look at the document properties
In your PDF viewer, open the document properties panel. Depending on the viewer you may see a "Tagged PDF" entry reading yes or no, along with the producer, the creation date and sometimes a conformance statement.
Treat a yes with some caution. It tells you tags are present, not that they are correct. A document can be tagged throughout with every heading at the wrong level and every list built out of ordinary paragraphs, and the panel will still say yes. A no, on the other hand, is conclusive: there is nothing there at all.
The producer field is worth a glance too. It tells you which software wrote the file you are holding, which is often not the software it started life in.
Test 3: convert it to Word and read what comes out
This is the most revealing of the three. Convert the PDF to Word and open the result.
A structured document arrives recognisable. Headings are headings, lists are lists, tables are tables you can click into and edit. A document with no structure arrives as something else entirely: a mass of separate text boxes, paragraphs broken at every line, tables scattered into loose cells, images floating away from the text they belonged to.
One honest caveat. A poor conversion can occasionally be the converter’s fault rather than the file’s. The way to tell is to try the same file in a second tool. If it comes out badly everywhere, the file is the reason, and no converter will rescue it, because there is nothing in the document for a converter to follow.
This is also why the test is useful beyond the diagnosis. Conversion is a demand placed on the structure, and the structure either answers it or it does not.
What to do with the answer
If the three tests say your document is structured, you are in good shape. Keep the source file, and be aware that structure can still be lost later: operations that rewrite a PDF do not all preserve it.
If they say it is not, the most important thing to know is that you cannot realistically repair it by hand. Structure is created at export, not retrofitted afterwards. The fix is upstream:
- Go back to the original document, in Word, Google Docs, InDesign or whatever produced it, and use real heading styles, real lists and real tables rather than manual formatting. Then export to PDF with the tagging or accessibility option enabled. Nearly every modern authoring tool has one, usually switched off by default.
- If the file is a scan, start with OCR. Running
- Keep the structured source. The PDF is the output, not the master copy, and that distinction is what lets you produce a corrected version in minutes rather than rebuilding from the page.
If accessibility obligations apply to your documents, this is the same ground seen from the legal side, which we covered in what the law now expects of accessible PDFs.
Why this matters more than it used to
For twenty years this layer concerned one audience: people using assistive technology. It was treated as a compliance task, scheduled last and done thinly.
That calculation has changed. The same structure now decides whether a document reflows on a phone, whether a search engine understands its hierarchy, and whether an automated system quoting your document quotes the right passage under the right heading. One piece of work, several beneficiaries, only one of which is a legal obligation.
It is also why the question of how you verify any of this is worth taking seriously. We set out the technical version of this argument in our analysis published with the PDF Association, the reference body for PDF standards, of which PDFSmart is a Full Member. More on what that membership means, and what it does not, in how we build PDFSmart on the PDF standard.
FAQ (Frequently Asked Questions)
What is a tagged PDF?
A tagged PDF carries a hidden description of its own content alongside the visible page: which parts are headings, lists, tables and figures, and the order in which the document reads. Tags are what let software understand a document rather than merely display it.
How do I know whether my PDF is tagged?
Copy a table or a two-column passage and paste it into a plain text editor. If the result is coherent, the document carries usable structure. The document properties panel in most viewers also shows a "Tagged PDF" entry, though it confirms only that tags exist, not that they are correct.
Why does my PDF convert to Word so badly?
Usually because the PDF has no logical structure for the converter to follow, so it has to infer paragraphs and tables from visual position alone. If the same file converts badly in more than one tool, the file is the cause rather than the converter.
Does OCR add structure to a scanned PDF?
No. OCR produces a selectable, searchable text layer over the image, which is a necessary first step for any scan. It does not create logical structure: it will not mark which line was a heading or which block was a table.
Does a PDF have to be tagged to be accessible?
In practice, yes. Assistive technology relies on the structure layer to navigate a document, announce headings and read tables in a sensible order. A document without it can be read aloud word by word but cannot really be navigated.
Can structure be added to an existing PDF?
Specialist remediation tools exist, but the work is slow and the result is rarely as good as a correct export. For most documents it is faster and more reliable to go back to the source file, apply real styles, and export again with tagging enabled.