Foundations
Accessibility review rests on five bodies of knowledge. A review that omits one of them will still produce findings, but predictably narrow ones.
They are ordered below by how much of the work rests on each. The first three determine what a reviewer is able to see. The last two determine whether what they find can be acted on by anyone.
Disability and assistive technology
A barrier is produced by the relationship between a person and an environment, not by a diagnosis. Reviewing well requires knowing the major categories of disability, how screen readers, magnification, speech input and switch access interact with content, and how temporary and situational constraints produce the same barriers.
Accessibility and universal design
Accommodation resolves one person’s access need after the fact. Design anticipates variability in advance. Both remain necessary, and the distinction is what tells a reviewer whether a finding calls for a repair, a redesign, or a change to a process.
Web and digital accessibility standards
WCAG organises digital accessibility around four principles: perceivable, operable, understandable and robust. Fluency means interpreting a success criterion against a specific piece of content, which is a different skill from running a checker across it.
Learning design and Universal Design for Learning
Universal Design for Learning asks whether a learning experience offers multiple means of engagement, representation, and action and expression. A course can satisfy every technical criterion and still narrow who is able to take part, which is what this lens is for.
Law, regulation and institutional governance
Findings change practice only where someone holds the authority to act on them. That means knowing which obligations apply to the organisation, and where accessibility is settled in practice: procurement, platform selection, content workflows and maintenance.
One review, three lenses. WCAG examines the artefact. Universal Design for Learning examines the learning experience. Inclusive design examines the system that produced both. The three answer different questions, and a finding is only complete once it names which of the three it belongs to, because that is what determines who is able to resolve it.
Process
Six stages, in the order they are carried out. Few reviews run all six; naming them all is what makes visible what an abbreviated review leaves out.
Scope and depth vary with context and resources, and stages three, four and five can be reordered or run in parallel. Stages one and six hold the rest together: without scoping, findings do not connect to the organisation, and without reporting, they do not reach anyone able to act.
Stage 1 Establish scope and context
Understanding the environment before assessing it
The reviewer establishes which content and platforms are in scope, who the learners are, what is already known about the barriers they encounter, and which accessibility obligations apply to the organisation. The scope is agreed in writing before assessment begins.
The judgement call is how much to take in. A scope wide enough to be representative and narrow enough to examine properly is what separates a review that can be acted on from one that returns a list nobody can prioritise. Where testing every page is not feasible, a documented sampling approach, such as the one set out in the W3C evaluation methodology, keeps the basis of the findings checkable.
- A written scope, recording what was excluded and why
- Learner context: known barriers and assistive technologies in use
- The standard, version and conformance level the review will apply
- A prioritised inventory of content and systems
Stage 2 Analyse barriers
Identifying what blocks participation, and where it lives
Barriers are examined across physical and motor, cognitive, sensory and situational dimensions. Two distinctions carry most of the weight: whether an issue blocks a task or adds friction to it, and where the barrier actually lives.
A barrier in content is remediated. A barrier in a platform requires a configuration change or vendor engagement. A barrier in a workflow or a policy requires a structural response that no individual designer can make alone. A review confined to content will report the first category accurately and leave the other two unreported.
- A barrier inventory categorised by type and severity
- Task-blocking barriers separated from usability friction
- Each barrier attributed to content, platform, workflow or policy
Stage 3 Audit against the standard
Automated and manual testing, and the difference between them
Automated tools detect a defined subset of issues quickly and consistently, among them missing alternative text, heading structure, colour contrast and ARIA errors. They are an efficient first pass and a poor last one.
Manual testing supplies what automation cannot: keyboard-only navigation, screen reader behaviour, magnification and reflow, and the judgement of whether alternative text conveys what an image is doing. The W3C states that no tool alone can determine whether content meets accessibility standards, so results are reported as findings that require interpretation rather than as a pass or a failure.
- Findings mapped to specific success criteria and content types
- Keyboard and screen reader results, with the environments tested named
- Contrast, reflow and text-spacing findings
- An explicit record of what automated testing could not determine
Stage 4 Evaluate with users
Testing with people who encounter the barriers
Structured evaluation with learners who use assistive technologies surfaces the distance between content that conforms and content that works. Task-based sessions focused on specific tasks or interface elements generally yield more usable findings than broad, undifferentiated studies.
Participation is consented, compensated and scoped. Participants are asked to complete tasks, not to audit the product; interpreting what happens remains the reviewer’s responsibility, and a small sample is reported as a small sample.
- Task completion findings, with points of friction located
- Heuristic evaluation from a disability-centred perspective
- Documented participant feedback and its analysis
- Usability barriers not visible to a technical audit
Stage 5 Assess the learning design
Examining the experience, not only the artefact
A technically conformant course can still narrow who is able to participate. This stage examines whether the design offers genuine alternatives across engagement, representation, and action and expression, and whether those alternatives are equivalent rather than nominal.
It also examines assessment. The question is whether the format of a task measures the outcome it is intended to measure, or whether it additionally measures something unrelated to that outcome: speed, recall, a particular sensory channel, or access to a specific device.
- An alignment map across the three UDL principles
- Assessment findings identifying format-based barriers
- Recommendations for flexible pathways and equivalent alternatives
Stage 6 Report and plan remediation
Findings in a form the organisation can act on
Findings are translated into a prioritised remediation plan that distinguishes what can be corrected quickly from what requires structural change, and connects each item to the workflow, policy or contract where it will be resolved.
Different audiences need different documents. Technical teams need specific and reproducible detail. Designers need principles and concrete alternatives. Those who control budget and procurement need prioritisation, resourcing and regulatory context. A single report written for all three usually serves none of them well.
- A prioritised remediation plan stating effort and impact
- Separate outputs for technical, design and decision-making audiences
- Findings connected to governance, procurement and content workflows
- A schedule for re-review
Decisions
Six choices that determine what a review will produce. Each can reasonably be answered more than one way, and each answer has a cost.
These are settled at scoping, and each one narrows what the review is able to report. Naming them in advance is what allows the person commissioning a review to recognise afterwards whether they received the review they needed.
Scope
What exactly is being reviewed, and what is being left out?Scope runs from a single course, document or platform feature, through a full programme or credential, to organisation-wide systems and policy. A narrow scope produces findings that are specific and immediately actionable but may miss the pattern that generated them. A wide scope locates the pattern but rarely reaches the depth needed to resolve any single instance. Whichever is chosen, recording what was excluded is what keeps the findings honest.
Standard
Which standard, at which conformance level, is the review conducted against?WCAG 2.2 Level AA is the common baseline for digital content. Organisations subject to federal regulation in Canada are moving toward CAN/ASC-EN 301 549, which incorporates WCAG and adds requirements covering hardware, software and documents. Naming the standard and version at the outset determines what counts as a finding; changing it partway through makes the results incomparable.
Depth
How thorough does this review need to be to answer the question being asked?A rapid assessment identifies major barriers for prioritisation. A comprehensive audit combines technical testing, user evaluation and learning design assessment. A continuous review embeds checks in the design cycle. Rapid assessment trades completeness for speed, which is a defensible trade where resources are constrained, provided its findings are treated as a starting point rather than a full picture.
Participation
Are the people who encounter the barriers part of the evaluation, or only its subject?Expert review alone is faster and less expensive. Structured user testing surfaces barriers that conformance testing does not reach. Co-design extends participation into deciding what is changed and how. Each step increases both what the review is able to find and what it costs in time, coordination and compensation.
Timing
Where in the design cycle does the review sit?Review before development shapes decisions before anything is built. Review before launch catches barriers while changes remain inexpensive. Remediation after launch addresses content already in use, where learners are already encountering the barriers and the cost of change is highest. Many organisations begin with the third and move earlier over time.
Audience
Who has to act on the findings, and what do they need in order to do so?A report written for developers, one written for learning designers and one written for those who hold budget and procurement authority differ in detail, framing and length. Settling the audience before the review begins changes what evidence is gathered, not only how it is presented afterwards.
Competencies
What a person conducting an accessibility review must be able to do. Each capability maps onto a stage of the process above.
These are capabilities rather than topics, so each can be observed in the work produced. They are drawn from the professional bodies of knowledge named in Sources and from the process set out in Process.
Analyse barriers and rank them by what they block
Produces an inventory in which task-blocking barriers are separated from friction, and each item is attributed to content, platform, workflow or policy. Without this, every finding arrives at the same priority.
Apply a standard to specific content
Reads a success criterion against a particular piece of content and states whether it is met, on what basis, and what would meet it. Requires working knowledge of WCAG and, for regulated organisations, of the ICT standard that incorporates it.
Test with assistive technology
Navigates content by keyboard alone and operates at least one screen reader and one magnification tool competently enough to tell a defect in the content from unfamiliarity with the tool.
Evaluate a learning experience against UDL
Identifies where a course offers only one path through engagement, representation or expression, and proposes alternatives that are equivalent rather than nominal.
Examine assessment for barriers unrelated to the outcome
Distinguishes what a task is intended to measure from what its format additionally requires, and proposes formats that preserve the first without imposing the second.
Specify remediation, not only detect defects
Writes fixes that a developer, designer or instructor can implement without further interpretation, across documents, media, HTML, LMS content and assessments.
Conduct evaluation with the people affected
Plans and facilitates sessions with participants who use assistive technologies, under informed consent and fair compensation, and analyses what is observed without overstating a small sample.
Trace accessibility through governance and procurement
Locates where in an organisation a barrier was introduced and who is able to prevent its recurrence — a contract, a template, a workflow or a maintenance schedule — and addresses the finding to that point.
Outcomes
What a review produces inside a project, and what tends to follow over a longer period. The two carry different evidential weight.
Outcomes within the project are observable during the engagement and can be verified by someone else. Longer-term change is what the available evidence suggests tends to follow, and is stated here with the hedging that evidence supports.
Within the project
- Barriers are named and prioritisedAn inventory that distinguishes what blocks a task from what slows it, with each item attributed to the content, platform, workflow or policy where it originates.
- Common defects are correctedMissing captions and alternative text, insufficient contrast, unlabelled controls and inaccessible documents are among the barriers that can usually be resolved soon after they are identified.
- A conformance position is documentedA record of what was tested, against which standard and version, what was found and what remains outstanding. Federal regulation in Canada will require conformity assessments of this kind from federally regulated entities beginning in 2027.
- Team capability growsPeople who take part in a review acquire practical accessibility literacy that carries into the next piece of content they build.
Over time
- Accessibility moves earlier in the design cycleWhere review is repeated, barriers tend to be addressed at the design stage rather than after publication, which is the point at which changes are least costly to make.
- Benefits reach beyond learners with disabilitiesClearer structure, consistent navigation and alternative formats are used by learners who have disclosed no disability. Meta-analytic evidence for UDL reports positive effects on educational outcomes alongside considerable heterogeneity between studies, so the finding is best treated as directional rather than precise.
- Procurement and governance changeOrganisations that have been through review are better placed to state accessibility requirements in vendor contracts and to evaluate the responses they receive.
- Review becomes periodic rather than singularBarriers accumulate as content is added and platforms are updated. Where review is scheduled rather than commissioned in response to a complaint, the backlog is smaller at each pass.
What a review does not do. It identifies barriers; it does not remove them. Conformance with a technical standard does not establish that a learning experience is usable, and the W3C is explicit that no tool can determine conformance on its own. Findings describe content as it stood on the date of testing, in the environments named, and they age as content and platforms change. A review also cannot supply the authority to act on what it finds; where that authority is absent, the report will document barriers that persist.
Reflection
Four questions about current practice, and four descriptions of what different patterns of answer usually indicate.
Choose the option closest to the current situation rather than the one that would be preferable to report. The descriptions that follow address the situation, not the people working within it.
Questions
-
How is accessibility review currently carried out in your organisation?
Consider what happens routinely rather than what happened once.
- No formal review takes place
- Automated scans are run when an issue is raised
- Periodic reviews combine automated and manual methods
- Review is built into the content development cycle
-
How well understood are the accessibility obligations that apply to your organisation?
Obligations differ by jurisdiction, sector and organisation size.
- The applicable legislation has not been identified
- The legislation is known but the obligations are unclear
- The obligations are understood and documented
- Regulatory developments are monitored and practice is adjusted
-
To what extent do learners with disabilities inform this work?
This asks about structured participation, not about handling complaints.
- Learners with disabilities are not involved in review
- Feedback is gathered informally when issues are reported
- Structured testing is conducted with learners with disabilities
- Learners with disabilities help decide what is changed
-
Where does responsibility for accessibility sit?
Consider where decisions are made rather than where the interest lies.
- Responsibility is not formally assigned
- Responsibility rests with one or two individuals
- Responsibility is shared across a team with defined roles
- Responsibility is held in governance, procurement and design processes
Reading your answers
Take the option you chose most often. If your answers are evenly split, read the lower of the two bands.
No established review practice
In this pattern, barriers are being identified by learners rather than by the organisation, and usually arrive as a complaint or an accommodation request. Nothing about that reflects on the people doing the work; it reflects the absence of a process for finding barriers before learners do. The most useful next step is a single scoped review of one high-traffic course or platform area, conducted against WCAG 2.2 Level AA, to establish what the barriers actually are before deciding how to organise the work.
Review is reactive
Checks are happening, but they are triggered by problems rather than scheduled, and automated results are carrying more weight than they can bear. This is the most common position, and it produces a reliable picture of a narrow band of issues. The next step is to add manual testing — keyboard navigation and one screen reader — to whatever is already being scanned, which will show how much the automated pass leaves unreported.
Review is periodic and structured
Method is established and findings are produced consistently, so the constraint is usually reach rather than rigour: review is thorough where it happens and absent where it does not. The next step is to move one review earlier in the design cycle, so that at least one piece of content is examined before publication rather than after.
Review is embedded
Accessibility is settled where content and systems are settled, and review is generating findings that reach people able to act on them. The remaining work is usually in maintenance and procurement, where new barriers enter. The next step is to require a conformity assessment as a condition of the next platform or content purchase, so that barriers are not acquired faster than existing ones are removed.
Sources
Every claim in this resource is traceable to one of the sources below. They are grouped by the foundation each supports, in the order of Section one.
Each entry carries a label naming what kind of source it is, so a reader can tell a standard from a study from a professional body of knowledge. Each address is public and stable at the time of the review date in the footer.
Disability and assistive technology
International Association of Accessibility Professionals. Certified Professional in Accessibility Core Competencies (CPACC) Body of Knowledge. Used here for models of disability, disability categories, assistive technologies, and the distinction between accommodation and design.
Accessibility and universal design
Inclusive Design Research Centre, OCAD University. Used here for the framing of disability as a mismatch between a person and a design, and for the principle that a design change acts on the wider system it sits in.
Inclusive Design Research Centre, OCAD University — idrc.ocadu.ca
Web and digital accessibility standards
World Wide Web Consortium. Web Content Accessibility Guidelines (WCAG) 2.2. W3C Recommendation, 5 October 2023. The technical baseline referred to throughout this resource.
Web Content Accessibility Guidelines (WCAG) 2.2 — w3.org/TR/WCAG22
World Wide Web Consortium. Website Accessibility Conformance Evaluation Methodology (WCAG-EM) 1.0. W3C Working Group Note. Used here for scoping, sampling and reporting in Stage one.
Website Accessibility Conformance Evaluation Methodology (WCAG-EM) 1.0 — w3.org/WAI/test-evaluate
W3C Web Accessibility Initiative. Selecting Web Accessibility Evaluation Tools. The source of the statement that evaluation tools cannot determine accessibility on their own and can only assist in doing so.
Selecting Web Accessibility Evaluation Tools — w3.org/WAI/test-evaluate/tools/selecting
Accessibility Standards Canada. CAN/ASC-EN 301 549:2024, Accessibility requirements for ICT products and services. National Standard of Canada, adopted 2024 and available at no charge. Incorporates WCAG and extends to hardware, software and documents.
International Association of Accessibility Professionals. Web Accessibility Specialist (WAS) Body of Knowledge. Used here for technical audit method.
Web Accessibility Specialist (WAS) Body of Knowledge — accessibilityassociation.org
WebAIM. The WebAIM Million: the 2026 report on the accessibility of the top 1,000,000 home pages. Published March 2026. Automated testing detected WCAG 2 failures on 95.9% of home pages, at an average of 56.1 errors per page; WebAIM notes these are only the errors automated testing can detect.
Learning design and Universal Design for Learning
CAST. Universal Design for Learning Guidelines, version 3.0, released July 2024. The source of the three principles used in Stage five.
Universal Design for Learning Guidelines — udlguidelines.cast.org
Almeqdad, Q. I., Alodat, A. M., Alquraan, M. F., Mohaidat, M. A., & Al-Makhzoomy, A. K. "The effectiveness of universal design for learning: a systematic review of the literature and meta-analysis." Cogent Education, 10(1), article 2218191, 2023. Reports positive effects across thirteen studies alongside considerable heterogeneity, which is why the outcome above is stated as directional.
Law, regulation and institutional governance
Parliament of Canada. Accessible Canada Act, S.C. 2019, c. 10. Establishes the framework for a barrier-free Canada by 2040 across federally regulated sectors, with information and communication technologies among its priority areas.
Government of Canada. Regulations Amending the Accessible Canada Regulations (the Digital Technologies Accessibility Regulations), SOR/2025-255, registered 5 December 2025. Adopts CAN/ASC-EN 301 549 and adds obligations for conformity assessment, accessibility statements, procurement and training. They apply to federal public-sector entities from 5 December 2027 and to large federally regulated private-sector entities from 5 December 2028.
Regulations Amending the Accessible Canada Regulations — gazette.gc.ca
Legislative Assembly of British Columbia. Accessible British Columbia Act, S.B.C. 2021, c. 19. Received Royal Assent in June 2021. An independent review of the Act was initiated in March 2026 and is scheduled to report in 2027.
Province of British Columbia. Accessible British Columbia Regulation, B.C. Reg. 105/2022, in force 1 September 2022. More than 750 prescribed public sector organisations, including school districts, must maintain an accessibility committee, an accessibility plan and a mechanism for public feedback. The Regulation does not currently specify a technical web accessibility standard.
Every source above is publicly available at the address given. Legislation and standards are revised: the date this resource was last reviewed appears in the footer, and anyone acting on a regulatory question should confirm the current position with the issuing body.