By Cally Mervine Kiser
I have a complicated relationship with eLearning quality checklists.
Not because they are wrong, exactly. Most of them cover legitimate ground: Objectives are measurable. Assessment items match the objectives. Narration is accurate. Navigation works. Accessibility requirements are met. These things matter, and they should be on a checklist somewhere.
The problem is that a course can pass every single item on most quality checklists and still be genuinely difficult to learn from. I have seen it happen more times than I can count. The checklist gets a green light. The course goes live. The completion rates are acceptable. And the behavior we were trying to change does not change.
For years, I assumed that was a content problem: Wrong objective. Wrong audience analysis. Wrong level of detail. So I would go back to the content and poke at it and usually find that the content was fine. Accurate, relevant, appropriately scoped. The problem wasn’t what the course was saying.
Identifying the Problem
It took me an embarrassingly long time to ask the obvious follow-up question: If the content is fine, what else could be getting in the way?
The answer, it turns out, has been sitting in educational psychology research for decades. Working memory is finite. When learners engage with training, they are managing three competing cognitive demands simultaneously:
- The complexity of the subject matter
- The effortful processing that builds understanding
- The mental effort generated by how the content is presented, independent of the content itself
That third one is called extraneous cognitive load, and it is the one we almost never explicitly design against. We design for accuracy. We design for alignment. We design for engagement, or at least what we hope will feel like engagement. We almost never sit down and ask, “How much unnecessary mental effort is this design generating, and how do we reduce it?”
Reducing Extraneous Load
I did not have a systematic way to answer that question until I started spending more time researching multimedia learning principles. What I found there were four concepts so well-supported by experimental evidence, across so many studies and populations, that they have essentially become bedrock: coherence, signaling, redundancy, and segmenting. Each one describes a specific design pattern that either adds extraneous load or reduces it.
And each one translates directly into an audit question you can ask about a screen you have already built. That is the part I want to be honest about, because the word “research-based” gets thrown around so casually in our field that it has stopped meaning much. These aren’t principles someone extrapolated from a single study or dressed up in scientific language to make a framework sound more credible.
A 2022 review of reviews synthesized 29 systematic reviews, 1,189 underlying studies, and more than 78,000 participants. The effects are consistent across subject areas, age groups, and learning contexts. This is about as replicated as findings get in educational psychology.
So I built the checklist that wasn’t part of any of the other checklists.
4 Questions
One screen at a time:
- Coherence asks whether every element on the screen earns its place relative to the learning objective. Not whether it’s interesting or relevant in a general sense but whether it serves this objective on this screen.
- Signaling asks whether the structure is visible, whether a learner could glance at the screen and immediately understand what matters most.
- Redundancy asks whether narration and on-screen text are doing the same job or different jobs.
- Segmenting asks whether the content is paced in a way that gives the learner control.
Each question gets a score of zero, one, or two. Zero means the principle is clearly violated. One means it’s inconsistently applied. Two means the screen gets it right. Eight points total.
- A screen that scores between zero and three is a candidate for a real redesign pass.
- A screen that scores six to eight is in good shape and probably just needs small refinements.
- The middle range usually means one or two principles are doing most of the damage—and fixing those is faster than it sounds.
I want to be specific about that last point because it is the one that surprised me most when I started applying this systematically. The fixes are rarely about adding something. They are almost always about removing something, splitting something into two pieces, or changing which channel carries which information. No new graphics budget. No new narration recording session. No meeting with the SME about content changes.
The content stays. The design gets out of its own way.
Eliminating the ‘Design Tax’
There is one more dimension to this that I did not expect to care about as much as I do. A 2024 study found that neurodivergent learners reported significantly higher extraneous cognitive load than neurotypical learners in online learning environments, even when the inherent difficulty of the content was identical for both groups.
The design was charging neurodivergent learners a higher tax for the same material.
We don’t always know which learners in our audience are carrying that extra burden. We do know that reducing extraneous load helps everyone and helps some people more than others.
That’s not a minor footnote. That’s a reason to treat this as an equity issue, not just an efficiency issue.
The checklist exists now. It fits on one page. It takes about five to eight minutes per screen. And it finds things that no other quality review I have used has surfaced, because it asks a question none of those reviews were designed to ask.
Not: Is this content accurate? But: Is this design making the content harder to learn than it needs to be?
Those are different questions. And for a lot of the training sitting in your course library right now, the answer to the second one is yes. We can fix this. And in most cases, we can fix it faster than you think.
Explore Extraneous Load—& More—at the Evidence-Informed Design Online Conference
Don’t miss the Evidence-Informed Design online conference, October 14–15, where you will join Cally Mervine Kiser to learn why reducing extraneous load is essential and how to avoid the most common design mistakes. Keynote Nidhi Sachdeva will explore cognitive architecture as a foundation for effective design. In other sessions, you’ll learn to write clearer narration, build reflection and practice into learning, and use feedback effectively. Register now; it’s free if you are a Professional or Enterprise member!
Image credit: danijelala
