Making a complex LMS easier to work in.

Research into how faculty actually use Canvas, turned into a restructured information architecture and a redesigned dashboard. A month later the same findings became the starting point for building an LMS from scratch.

Canvas had the functionality faculty needed. The challenge was finding the right path through it: the structure made common tasks harder to find, understand and complete.

Canvas is one of the most widely used LMSs in higher education, but on the faculty side there was friction everywhere: navigation, course management, grading, finding content.

The point wasn't to make Canvas look better. It was to understand where the experience broke down, and what a better structure could look like.

Seven faculty interviews, diary studies and a heuristic review of course setup kept surfacing the same problems:

• Important functions were hard to find. No search or filters, and details like course settings and assignment status weren't visible where they were needed.
• Common workflows took trial and error. No templates or guidance for building courses and modules.
• Nothing had a clear hierarchy. The core tasks competed with everything else on screen.
• Grading was manual and slow. Time that should have gone to teaching went to admin.

One diary-study participant put it more bluntly:

I reorganised the experience around the way faculty think about their work, rather than the way the existing product grouped its features.

The proposed structure was tested against how people naturally grouped tasks and content.

Not everything could be fixed at once. The research decided what went first: everyday workflows, ahead of anything that added complexity without solving a core problem.

With the structure settled, a common faculty job (setting up a course, then grading it) became one continuous path with fewer decisions along the way: 14 interactions from start to published grades.

This is where the structural decisions became interface decisions: what sits on the dashboard, what's one click away, and what can wait.

A redesign this size needed a consistent system, not a set of one-off screens.

The dashboard and its patterns were tested and refined around discoverability and hierarchy.

An attention heat map scored the layout 45 for clarity and 66 for focus, and flagged weak hierarchy in places and inconsistent card padding. Both went into the final pass.

A month later, I joined Unison Flow to help build an LMS from scratch. The Canvas research became the brief.

Canvas research → a clearer information architecture → product principles → a four-product ecosystem on one design language → an LMS MVP.

Research is worth more when it changes what gets built.

The Canvas redesign showed me the structural problems inside an LMS. Unison Flow taught me how to carry those findings into product decisions, a design system and a real 0 to 1 build.