Accessible Documents Should Start with Content Design, Not End with PDF Remediation

Aug 26

7 min read

Updated: Sep 2

BlindSpot Insights Accessibility and Universal Design branded article cover.

Accessible documents should be usable from the moment they are created, not repaired after the writing, design and approval work is finished. Yet many organisations still treat accessibility as a technical step performed on the final PDF, when the structure, language, layout and format decisions that determine the experience were made much earlier.

This matters because a document is rarely just a file. It may explain a service, ask someone to make a decision, provide instructions, communicate an entitlement or record information that affects a person’s work. If the document cannot be navigated, understood or adapted using assistive technology, the barrier sits directly between the person and the outcome they need.

Why accessible documents become a remediation problem

Document accessibility often arrives late because responsibility is attached to publishing rather than authorship. The content team writes the words, a designer creates the layout, subject matter experts add changes, executives approve the final version and only then does someone ask whether the PDF is accessible. By that stage, accessibility is expected to fit around decisions that have already become expensive or politically difficult to change.

Late remediation can add tags, adjust reading order and repair some missing descriptions, but it cannot easily simplify a confused argument, replace an unnecessary table or resolve a document that was designed primarily as a visual artefact. The remediation specialist is left correcting symptoms while the source of the problem remains in the workflow.

The result is familiar: accessibility is described as slow, costly and complicated, even though much of that cost was created by postponing it. A document designed with structure and user needs in mind is generally easier to publish, maintain and convert into different formats. Accessibility becomes harder when the organisation first removes the structure and then pays someone to reconstruct it.

PDF accessibility is not a final checkbox

PDF accessibility requires more than a successful result from an automated checker. A usable PDF needs meaningful headings, lists and tables, a logical reading order, descriptive links, appropriate alternative text, sufficient contrast and document properties that help people identify and navigate the content. Forms add further requirements, including labels, instructions, keyboard operation and understandable error handling.

The Australian Government Style Manual advises organisations to create webpages by default and use PDFs only where research identifies a genuine need for the format. It highlights recurring problems with scaling, navigation, search, maintenance and assistive technology. The Australian Human Rights Commission also notes that PDFs cannot be made fully accessible in every context, particularly on mobile devices, and points to alternative formats such as EPUB.

This does not mean every PDF should disappear. Contracts, printable forms, signed records and documents with a fixed presentation may have legitimate reasons to exist. It does mean that choosing PDF should be a user need decision rather than an organisational habit, and that publishing a tagged PDF should not prevent the same information being offered in a more flexible format.

Content design determines structure before software does

Content design begins with the task the reader is trying to complete. It asks what information people need, in what order, with which terms explained and what action should follow. These decisions improve accessibility because they reduce the cognitive work required to interpret the document before any technical features are added.

Clear headings allow a screen reader user to scan sections rather than listening from the beginning. Short, descriptive links make sense when read out of context. Plain language reduces ambiguity for people with cognitive disability, people reading in a second language and anyone working under time pressure. Well designed instructions separate prerequisites, actions and consequences so the reader can build an accurate mental model.

Software can mark text as a heading, but it cannot decide whether the heading reflects the reader’s question. It can identify a long paragraph, but it cannot determine whether the argument should be reorganised. Accessible documents therefore depend on editorial judgement as much as technical conformance.

HTML should be the default when the need is information

When the primary need is to read, understand or act on information, HTML usually offers a stronger starting point than PDF. It can reflow across screen sizes, respond to browser and operating system settings, support direct links to sections and remain easier to search and update. People can enlarge text or adjust colours without fighting a page that was designed around a fixed sheet of paper.

HTML also supports a more reliable relationship between visible structure and programmatic structure. Headings, lists, landmarks, tables and form controls can be expressed semantically and tested in the environment where people will use them. A printable stylesheet can meet many of the reasons organisations give for creating a PDF while preserving the flexibility of the web.

The format decision should still follow evidence. Some audiences may need an offline file or a controlled layout, and some processes may require a downloadable record. The better practice is to identify those needs explicitly, provide an accessible alternative where necessary and avoid making one format carry every purpose.

Source files carry accessibility forward

A well structured source document is the most valuable asset in an accessible publishing process. Styles should be used for headings rather than enlarged bold text. Lists should be created using list tools rather than typed symbols. Tables should be reserved for data, given clear headers and kept as simple as the information allows. Images should have alternative text or be marked decorative when they add no information.

These choices preserve meaning when content moves between Word, presentation software, content management systems and PDF. They also make future updates safer because the next author inherits structure rather than a collection of visual conventions. Templates can help by making the accessible choice the easiest choice, but only when they are paired with guidance and review.

The source file also affects accountability. If only the remediated PDF is accessible, every later edit creates a risk that the accessibility work will be lost. Maintaining an accessible source allows the organisation to regenerate outputs without beginning again each time a phone number, date or policy detail changes.

Automated checkers can detect problems, but they cannot understand the experience

Automated tools are useful because they can find repeatable technical issues quickly. They may identify an untagged document, missing language settings, images without alternative text or headings that skip levels. Those findings make a good quality control layer, particularly when teams publish documents frequently.

A passing score is not evidence that the document is understandable or efficient to use. A checker cannot reliably decide whether alternative text communicates the purpose of a diagram, whether a table should have been a list, whether the reading order matches the intended meaning or whether the instructions make sense without visual positioning. It also cannot tell whether the chosen format creates unnecessary effort on a mobile device.

Human review must therefore include keyboard navigation, structured inspection and testing with relevant assistive technologies. Where a document supports an important or high risk process, involving people with disability provides evidence that conformance alone cannot supply. The question is not simply whether the file contains the expected technical properties, but whether people can complete the task.

Accessibility needs ownership across the document lifecycle

Accessible publishing works best when responsibility is distributed but clear. Authors own language and structure. Designers own visual clarity without disconnecting presentation from meaning. Subject matter experts protect accuracy while accepting that complexity should be explained. Publishers confirm format, metadata and technical quality. Leaders give teams enough time and authority to resolve barriers before approval.

Procurement also matters because authoring, design and document management tools shape what staff can produce. That does not make accessible documents a repeat of accessible procurement. The operational question here is what happens after the tool is purchased: whether templates, training, review gates and publishing channels consistently preserve accessible practice.

Governance should be proportionate. A two page internal update does not need the same assurance process as a public form that determines access to an essential service. A simple risk model can consider audience, consequence, complexity, publication life and likelihood of reuse, then match the review effort to the potential impact.

What an accessible document workflow includes

A practical workflow moves accessibility decisions towards the beginning and keeps them visible through approval. It does not require every employee to become a PDF specialist. It requires each role to understand the decisions it controls and the evidence needed before publication.

A defined user need. State who needs the information, what they must be able to do and why a document is the right way to support that task.
A deliberate format decision. Prefer HTML for information that needs to be read, searched, enlarged, updated or used across devices. Record the reason when PDF is necessary.
An accessible source. Use semantic headings, genuine lists, simple tables, descriptive links, useful alternative text and a logical content sequence before conversion.
A review matched to risk. Combine automated checks with human inspection and assistive technology testing, increasing assurance where barriers would have serious consequences.
An accessible alternative. Where the chosen format has unavoidable limitations, provide the same information through another usable format and make that option easy to find.
Clear ownership for updates. Keep the source accessible, assign responsibility for maintenance and repeat the necessary checks whenever the content changes.

The better test is whether people can use the information

The Australian Human Rights Commission’s 2025 guidance frames digital accessibility as equal access to goods, services and information. That is a more useful ambition than producing a file that appears compliant in a report. Equal access asks whether people can find the information, understand it, navigate it and complete the same task with comparable independence.

Accessible documents support more than assistive technology. Strong structure helps people skim. Plain language helps people decide. Flexible formats work better on small screens. Clear instructions reduce mistakes and support requests. The same practices that remove disability barriers often make information more resilient across devices, contexts and levels of digital confidence.

PDF remediation remains valuable when PDF is genuinely required, but it should be the final safeguard in a well designed process. The larger opportunity is to make accessibility part of content design, authorship, format selection and quality assurance. When that happens, organisations stop repairing documents at the end and start creating information people can use from the beginning.