The learners you exclude never tell you

The learners you exclude never tell you

Nobody complains.

That is the part that should stop you. When a course shuts someone out, you almost never hear about it. The learner who cannot use it does not fill in the feedback form. They do not email you. They drop out, or they battle through and fail, or they ask a mate to read it to them and never breathe a word of it to anyone.

Your completion data looks fine. Your feedback scores look fine. And somewhere in that group, someone was locked out of learning you made for them, and the whole system stayed perfectly quiet about it.

I have been designing learning for a long time, and if I am honest, accessibility is the thing our profession has been slowest to really build into how we work. Not out of malice. Out of defaults. That is how exclusion usually happens in course design. Almost nobody sits down and chooses it. We inherit it, from templates, from habits, from whatever the tool does when you leave it on its factory settings.

So I want to be specific about who gets left behind, and what the actual decisions are. Because accessibility in our world tends to get talked about either as a legal box to tick or as a warm feeling about inclusion, and neither of those helps you on a Tuesday afternoon when you are just trying to build a module and get it out the door.

The numbers are bigger than you think

More than one in five Australians now lives with disability. The Australian Bureau of Statistics put it at 21.4% in its most recent survey, up sharply from 17.7% just four years earlier. That is 5.5 million people. Not all of those affect how someone uses a course, but plenty do, and the ones that do are so often invisible.

Then there is literacy, which barely comes up in these conversations and really should. The most recent national assessment of Australian adult literacy found that around 44% of adults sit at Level 2 or below. In plain terms, they struggle with exactly the kind of dense, multi-step text most workplace training is written in. That figure is over a decade old now, and there has been no newer national release since, but nothing about how we write has made it likely to have improved. That is close to half the working-age population, and a far bigger group than most of us picture when we hear the word "accessibility."

Now add the learners with English as a second language. Add the ones on a phone, on a train, on a noisy warehouse floor. Add the ones having a hard week, a broken wrist, an ear infection, a migraine, a baby who was up all night.

Accessibility is not a niche accommodation for a handful of people. It is just an honest description of how most of your learners actually meet your course.

Where it actually goes wrong

Here is what I mean about defaults. These are the small decisions that quietly lock people out, and not one of them feels like a decision when you are making it.

Video with no captions. The most common one, and the one that costs the most. No captions shuts out Deaf and hard of hearing learners completely. It also shuts out anyone in a noisy room, anyone without headphones, anyone whose first language is not English and who needs to see the words to follow them, and anyone who simply takes text in better than speech. Captions were never just a Deaf accommodation. Given the choice, a huge chunk of your cohort watches with them on.

Drag and drop. Lovely in the authoring tool. Impossible for someone using only a keyboard, hard for anyone with a motor impairment, and a headache on plenty of touch screens. WCAG 2.2 even added a specific rule for this, asking for a pointer alternative, and most rapid-authoring output still misses it.

Colour doing all the work. Green for right, red for wrong. Around 8% of men have some form of colour blindness. If colour is the only thing carrying the meaning, you have made your feedback unreadable for a slice of the room, in a way they will never once mention to you.

Text trapped in an image. Screenshots of documents, infographics with the real substance baked into the picture, scanned PDFs. A screen reader gets nothing from any of it. So the learner gets nothing either.

The timed quiz. Most time limits are there for admin reasons, not learning ones. They quietly punish people who process more slowly, people using assistive tech, people who are anxious, people working in a second language. Unless speed is genuinely part of the skill you are assessing, that timer is measuring something you never meant to measure.

The wall of text. Not a technical WCAG failure, but this is where that literacy number bites hardest. Long paragraphs, formal tone, tangled sentences, jargon with no explanation. Every one of those is a barrier, and they stack on top of each other.

Autoplay, auto-advance, no way back. Content that moves on by itself takes control away from people who need to go at their own pace, and that is far more people than you would think.

None of this is malicious. Every bit of it is a default someone accepted because the deadline was tight and the tool made it the easy option.

Why "compliant" is not the same as "accessible"

This is the part that honestly took me years to properly understand.

You can build a course that sails through an automated accessibility checker and still leave people out. Industry testing consistently finds automated tools catch only around a third of accessibility issues. They pick up the mechanical stuff, missing alt text, contrast ratios, form labels. What they cannot tell you is that your alt text just says "image1.png", or, worse, that it describes the picture perfectly but completely misses the point the picture was making.

And WCAG itself, is a floor. It is not a description of good learning. A course can tick every box in WCAG 2.2 and still be cognitively exhausting, written at a reading level half your learners cannot reach, and structured so nobody can ever find anything twice.

The Australian Government asks agencies to meet WCAG 2.2, and it is absolutely the right benchmark to work to. But treating it as the finish line is the same mistake we make everywhere else in this job. Meeting the standard means you are not doing obvious harm. It does not mean the learning actually works.

Accessibility is a design question, not a compliance one. Get the design right and the compliance tends to follow. It very rarely happens the other way around.

What actually helps

Some of this is genuinely easy, and I want to be honest about which bits.

Caption everything. It is cheap now, near enough to free, and there is no good reason left not to. Auto-captions still need a human pass, especially for technical words, names and accents, but the cost has all but disappeared. Just do it, every time.

Give every interaction a keyboard path. Whatever the mouse can do, the keyboard has to be able to do too. This is the most common failure in interactive elearning, and it is usually a five-minute fix while you are designing and an absolute nightmare to bolt on afterwards.

Never let colour carry the meaning on its own. Add an icon, a label, a word. Right and wrong should still make sense in black and white.

Write for the learner you actually have, not the one in your head. Short sentences. Plain words. One idea to a paragraph. Explain the jargon the first time you use it. This is not dumbing anything down. It is the difference between someone learning and someone quietly closing the tab. Given what we know about literacy, it is probably the highest-impact accessibility decision on this whole list, and it costs you nothing but a bit of care.

Bin the timer unless the speed is the point. If you are not actually assessing how fast someone works, do not assess it by accident.

Hand control back. Let people pause, rewind, reread, and move at their own pace. Autoplay is a choice, not a law of physics.

Try your own course with a keyboard, just once. Before it ships, put the mouse down and see if you can finish it using only the keyboard. It takes ten minutes and the first time will genuinely rattle you. That one habit will catch more real-world exclusion than any automated scan ever will.

The part I find hardest

I want to finish on the thing that actually sits with me.

The learners you have shut out almost never tell you. No angry email. No line in the feedback. Exclusion in learning is quiet by design, because the people on the receiving end have usually spent their whole lives being told the difficulty was theirs to sort out. So they work around it. They ask a friend. They give up softly and blame themselves.

Which means all those silent completions and calm feedback forms are not proof your course is working. That silence looks exactly the same whether everything is fine or whether someone just walked away.

The only way to really know is to design as though those learners are already in the room, because they are. More than one in five with a disability. Nearly half the population reading below the level most of our training is pitched at. Every cohort, every time, whether they say a word or not.

We pour so much energy into engagement, into polish, into making learning feel delightful. And all of that matters. But it is worth almost nothing to the person who could not get through the door in the first place.

The door was always a design decision. It still is.

Sources: Australian Bureau of Statistics, Survey of Disability, Ageing and Carers 2022; Australian Bureau of Statistics, Programme for the International Assessment of Adult Competencies 2011-12; W3C Web Content Accessibility Guidelines 2.2; WebAIM Million report on automated accessibility detection.