Tag PDF for Accessibility: Tags and Order
On this page
To tag a PDF for accessibility is to write into the file what each piece of content is, so a screen reader can announce a heading as a heading and follow the text in the right order. The tag tree replaces guesswork with structure. This guide explains what a tag is, the tags you will meet most, how reading order follows the tree, and the two routes to getting there, plus why an automatic pass always needs a human look.
What a PDF tag actually is
A tag is a label attached to a run of content. It says “this is a level-two heading”, “this is a paragraph”, “this list has three items”, “this is a data table with these headers”. The tags are not visible on the page; they live in a structure tree alongside the drawing instructions.
That tree is what assistive technology reads. Without it, a screen reader has to infer structure from position and font size, and it often guesses wrong. With it, the reader knows the document’s shape and can move by heading, list and table. The tag set and its conformance requirements are documented at the PDF/UA Foundation.
The tags you will meet most often
You do not need to memorise the full PDF tag set, but the common ones cover almost every document.
| Tag | What it marks | Notes |
|---|---|---|
| H1 to H6 | Heading levels | The outline a screen reader can navigate by |
| P | A paragraph | The default body element |
| L, LI, Lbl, LBody | List, list item, label, list body | Nested lists use the same set |
| Figure | An image or graphic | Carries the alternative text |
| Table, TR, TH, TD | Table, row, header cell, data cell | Headers are what make a table readable |
| Link | A hyperlink | Carries the destination |
| Artifact | Decoration | Skipped by assistive technology |
The Artifact tag is the one people forget. Page numbers, running headers, borders and background images should be artefacts, because a screen reader announcing “page 4 of 30” in the middle of a sentence is noise. Marking decoration as an artefact is as much a part of tagging as marking the real content.
How reading order follows the tag tree
Reading order is not a property you set separately; it is the order of the tags in the tree. A screen reader reads the tree top to bottom, so if the tree is wrong, the reading is wrong, even if every tag is correct in isolation.
This is where multi-column layouts fail. A two-column page drawn left to right, top to bottom produces a tag tree in that geometric order, so a reader hears the first line of column one, then the first line of column two, then back. The fix is to restructure the tree so column one reads in full before column two. That is a judgement about the document’s intent, and it is why an automatic pass cannot finish the job alone.
The same problem appears with sidebars, pull quotes, captions and footnotes. Any element that sits beside the main flow rather than within it is a candidate for a reading-order error. For how a screen reader experiences the result, see screen reader PDF.
The two routes: export, or tag what you have
There are two honest ways to get a tagged PDF, and which one you can use depends on whether you control the source.
Export from the authoring tool. If you still have the Word, Google Docs or InDesign file, export it as a tagged PDF. The structure is written at creation, from the styles you already used, so headings and lists come through correctly. This is the best route and it is the cheapest. It only works when the source exists.
Tag a PDF you did not author. For scans, vendor files and inherited archives, there is no source to export from. You tag the finished file, either manually in a desktop tool or with a browser-based fixer. This is the route that matters for the folders coordinators actually receive.
The split is stark: most PDFs a public body publishes were not authored by the person publishing them, so the export route is closed and tagging the existing file is the only option.
Why automatic tagging needs review
An automatic pass reads the layout and infers structure. It is good at headings in a clean single-column document and poor at anything unusual. It cannot write meaningful alternative text, because it does not know what an image conveys. It cannot judge a complex table’s headers, because it does not understand the data. It cannot fix reading order in a multi-column page, because it follows geometry rather than intent.
So the automatic result is a first draft, and the review is the work. A good tool makes the review tractable by telling you what it inferred and what it could not fix. TagRight tags a PDF in the browser and returns exactly that: the heading outline it detected, the language it found, and a list of what still needs a human, such as alt text for images and complex tables. It does not claim to finish the job, and it cannot tag a scan with no text layer. For the full manual sequence, see remediate PDF, and for what a finished file should contain, accessible PDF.
A worked example: a housing authority receives a vendor’s 60-page annual report as a PDF with no tags. They upload it, get back a tagged copy with a sensible heading outline and a note that 14 images need alt text and two tables need header review. They set the 14 descriptions and fix the two tables, then validate. The automatic pass handled the 55 pages of body text; the human work was two hours on the parts that mattered.
TagRight is a remediation aid and a triage tool, not a PDF/UA certification, so verify anything that must pass an audit. If you want the wider comparison before choosing a route, see PDF accessibility without Acrobat; you can start with a free document, and the plans and pricing are below a desktop suite.
Frequently asked questions
What is a tag in a PDF?
A tag is a label that says what a piece of content is: a heading, a paragraph, a list item, a table cell, a figure. Tags form a tree, and the tree is what a screen reader follows instead of the page image.
How do I add tags to a PDF I did not create?
You need a remediation route. Acrobat Pro has an automatic tag pass plus a tags panel for correction, and browser-based tools can tag a PDF you upload. Free checkers only report, they do not add tags.
Does automatic tagging produce an accessible PDF?
Not on its own. An automatic pass infers structure from layout and gets the routine parts right, but reading order, meaningful alt text and complex tables need review. Treat the automatic result as a first draft.
What is the difference between tags and reading order?
Tags say what each element is. Reading order is the sequence in which the tags are read, which follows the tag tree. A file can have the right tags in the wrong order and still fail a screen reader.
Can a scanned PDF be tagged?
Not until it has a text layer. A scan is an image, so it must go through OCR first, and the recognised text usually needs correction before tagging means anything.