Accessible Learning Design Is Not a Checklist
If accessibility is still a late-stage checklist item in your process, you are not doing accessible learning design. You are doing remediation, and you are paying twice for it.
The pattern is consistent across every large organisation I work with. The content gets built. It goes to review. Somebody runs an automated checker a fortnight before launch. A list of contrast failures and missing alt text comes back, a designer patches them, and the programme ships with a compliance statement nobody quite believes.
That process satisfies an audit. However, it does not produce learning that works for the people it excluded.
Why does accessible learning design keep getting bolted on late?
Because it is owned by the wrong stage and measured by the wrong thing.
Accessibility gets treated as a quality-assurance concern, so it arrives when QA arrives — at the end. And it gets measured as a pass rate against a standard, which makes it feel like a test to survive rather than a design constraint to work within.
Design constraints are useful early and expensive late. For example, this is equally true of brand and of platform limits. That is true of brand, of tone, of platform limits, and it is true here. A colour palette chosen in week one costs nothing. The same palette corrected in week nine means re-exporting every asset.
What does it actually cost to fix it at the end?
More than teams expect, and rarely in the budget line where it shows up.
Retrofitting captions, re-recording narration to match revised on-screen text, rebuilding an interaction that depends on drag-and-drop, reflowing a layout that breaks at 200 per cent zoom — none of that is a tweak. Each one reopens work that was signed off. Therefore each one costs another review cycle and another round of stakeholder time.
Take one example that recurs constantly. A module uses drag-and-drop matching because it tests better than multiple choice. Late in the process someone notes it cannot be completed with a keyboard. The options are now all bad: rebuild the interaction, add a parallel accessible version and maintain two, or ship it with a note. Chosen at brief stage, the same learning objective could have used an interaction that works for everyone and costs the same to build.
The second cost is subtler and larger. In fact, it compounds. Every retrofit teaches the team that accessibility is somebody else's problem arriving late, which guarantees the next project repeats it.
The organisations that have stopped paying this tax did one thing differently. They moved accessibility from the review stage to the brief stage. Not a new tool, not more training. Just a different place in the process.And the retrofit rarely produces good learning. It produces compliant learning. A screen reader can now announce the content, but the experience was never designed for someone using one.
Is accessible learning design just a compliance obligation?
No, and framing it that way is why it keeps losing budget arguments.
Every accessibility decision is also a design decision that benefits people well outside the group it targets. Captions get used by everyone in an open-plan office. Clear heading structure helps anyone skimming on a phone. Sufficient contrast helps anyone on a laptop near a window. Plain language helps every learner working in their second language, which in most enterprise cohorts is a large minority.
Ultimately, design for the edge and the middle gets better too. That argument wins budget. "We might get sued" does not, and it positions the work as a cost rather than an improvement.
What changes if you move it to the brief?
Four things, and none of them require new software.
The colour palette gets checked before anything is designed with it. The interaction patterns get chosen from a set that already works with a keyboard. The narration script and the on-screen text get written together rather than reconciled afterwards. And the review checks whether the design intent was met, instead of discovering the intent was impossible.
That is a process change, not a technology change. It costs a conversation at kickoff and saves a fortnight at the end.
It also changes who is accountable. When accessibility lives in testing, the tester owns it and has no authority to change the design. When it lives in the brief, the designer owns it and has every authority. Ownership without authority is how good intentions produce bad outcomes, and it is the structural reason this keeps failing in organisations that genuinely care about it.
How do you tell whether your process is genuinely inclusive?
Ask when accessibility first appears in your project plan. If the first mention is a testing task, you have your answer.
Then ask a harder question. When your last programme failed an accessibility check, what changed? If the answer is "we fixed the findings", the process is unchanged and the next programme will fail the same way. If the answer is "we changed the template", something actually improved.
The teams getting this right are not more virtuous. They have simply stopped treating accessible learning design as a gate at the end and started treating it as a constraint at the start, which is the only place a constraint is cheap.
Three changes worth making first
Fix the template, not the module. Most accessibility failures repeat because they are baked into a starting template. Correct the contrast, heading structure and focus states there and every future build inherits it. Write narration and on-screen text as one task. Most caption and audio-description problems are created by writing them separately and reconciling them later. One writer, one pass, both outputs. Put accessible learning design into the brief template. Not a policy. One line: who in this audience will struggle with this format, and what does the design do about it? That question, asked at brief stage, catches more than any end-stage checker, because it forces a design response rather than a test result.None of this requires a specialist, a new platform, or a bigger budget. It requires moving the same work earlier, where it is cheap, and giving it to the person who can act on it. Accessible learning design done properly is not more expensive than doing it badly. It is considerably cheaper. It just moves the cost to a stage where nobody notices it, which is precisely why it is worth doing.
