The brief for the SMVMCH nursing quality dashboard arrived as a stack of photocopies: thirteen different paper worksheets that ward nurses filled in every shift — audits, checklists, incident counts, quality indicators. Somebody then re-entered all of it into spreadsheets, weeks later, by hand.
It would have been easy to digitise the paper — thirteen forms become thirteen screens, done. It also would have been wrong. The project only worked because we started by asking a different question: not "how do we put these forms on a screen" but "what were these forms trying to find out."
Shadow the workflow before you sketch it
Watching the worksheets being used changed the design before a single wireframe existed. Nurses fill forms standing up, mid-shift, between patients — so entry had to survive interruption and never lose a half-completed record. And the same data points appeared on multiple sheets, copied by hand, which is where errors crept in.
The single most important finding: nurses are not the dashboard’s audience. They are its authors. The audience is the quality team and hospital leadership. Two different users, two different needs, one system — enter fast and trust it saved, versus see patterns and act on them.
Thirteen forms, one information architecture
Laid side by side, the thirteen worksheets collapsed into a handful of underlying record types tracked across wards and shifts. That became the architecture: instead of thirteen digitised forms, a small set of unified entry flows that route data everywhere it used to be copied.
The best thing a dashboard can do for a nurse is ask for each fact exactly once.
"Learn once" as a design principle
Hospital staff rotate constantly, and training time is effectively zero. So every entry flow uses the same pattern — same layout, same controls, same save behaviour. Learn one worksheet and you have learned them all. The dashboard side follows the same rule: every indicator is presented the same way, with the same colour logic for on-track and needs-attention.
Calm is a feature in clinical software
- No decorative data-viz — every chart answers a question the quality team actually asks.
- Colour is reserved for meaning; alarm red appears only when something is genuinely wrong.
- Dense tables are allowed — clinical users prefer scannable density to friendly whitespace.
- Every number can be traced back to the entries behind it, because trust in a dashboard is trust in its data.
The result replaced thirteen worksheets with one system that nurses learn once and leadership actually reads. Nothing about it is flashy — and in a hospital, that is exactly what good design looks like.
