User experience design

Why it matters

Teachers and students spend hours a week inside a learning platform. Every unclear label, buried setting and lost draft takes time that was meant for teaching and learning. A friction that costs one teacher a minute costs every teacher that minute, every time they meet it. Nobody files a complaint about a minute. They just stop using the feature.

The usual case for design in education is badly made. The claim that better interfaces produce better learning has hardly been tested: a systematic map of usability research on learning management systems found no studies linking a usability improvement to learning. The most quoted figure about the cost of fixing problems late traces back to course notes, and the study it is credited to cannot be found. A case built on numbers like these collapses the first time someone checks, and takes design’s credibility with it.

The instincts the field imports from consumer products can also point the wrong way. In one careful study, physics students taught through active learning learned more than peers in well-delivered lectures, and believed they had learned less. A team that tunes a learning product for satisfaction can end up tuning it against the learning.

What I have come to believe

Technology’s average effect on learning is modest. The spread around that average is wide, from products that changed nothing to products that moved learning a long way, and the same medium turns up at both ends. Hold the medium constant and vary the design of the learning, and the effects become meaningful. So the case I make for design is about that spread. Design is much of what decides which end a product lands at.

There are two kinds of effort, and they need opposite treatment. Interface friction costs attention and produces no thought: hunting for a control, holding a code in memory between screens and recovering from an error the system caused. Cognitive effort is the thought: recalling, predicting, explaining, getting it wrong and seeing why. Strip the first. Protect the second. Making text harder to read adds study time without any reliable gain in learning, so difficulty has to be the right kind.

No design is good in general. The same support that helps a novice can hold back someone who already knows the material, and the direction of the effect flips with prior knowledge. A design is good for a particular learner at a particular moment, which makes knowing who is using it, and when, part of the design.

At the scale of a national learning platform, a default is a rule. It binds people who will never meet the person who set it. That makes design a form of governance, and it is why accessibility, consistency and plain language belong in a shared system with principles, instead of being left for each team to rediscover.

And much of the experience is settled before any screen exists: in what gets bought, what the data model can hold and what policy requires. Design that arrives only at the interface is left to make a finished decision look good.

How I approach it

I start with the people who live in the product, and I follow them through whole tasks, end to end, instead of through the screens one team owns. The worst problems tend to sit at the joins, where one team’s screen hands over to another’s.

Then I sort the effort the product asks for. Each piece of effort gets one question: does it cost attention and produce no thought, or is it the effort the learning needs? The first gets designed out. The second gets kept and explained, because sooner or later someone will report it as a usability problem.

And I check what the team measures. A measure of satisfaction or time on task will pull the product towards ease. So I try to pair every measure of how the product felt with a measure of what the learner can now do.