Nurse schedulers were building rosters in a system that optimised for coverage alone. Staff experienced it as arbitrary and unfair, and the hospital was losing nurses to competitors with better scheduling.
STUDY 16 / 30 · ANONYMISED · NDA · ENTERPRISE INTERNAL · US · 42,000 EMPLOYEES
ENTERPRISE INTERNAL · US Shift scheduling that respects the person being scheduled.
RoleLead designer + researcher
Timeline16 weeks
Team2 designers, 1 researcher, 1 PM, 3 frontend
VerticalWorkforce management · healthcare system
Made fairness a visible, computed property of a roster rather than an outcome nobody could see, and gave staff a swap marketplace that did not require a manager.
Voluntary turnover fell measurably in the first two quarters. The scheduling system stopped appearing in exit interviews.
-22%VOLUNTARY TURNOVERfirst two quarters
-41%UNFILLED SHIFTSvs baseline
84%SWAPS WITHOUT MANAGERof all swaps
THE ARGUMENT
Why the obvious solution was wrong.
The study matters because the product problem was reframed before the interface was polished.
Scheduling software is written as a constraint-satisfaction problem - cover every shift, respect the rules, minimise cost. That framing is complete from the hospital's side and invisible from the nurse's. A roster that is optimal by that definition can still allocate every weekend in a quarter to the same person, and the system has no representation of that being a problem because fairness was never a variable.
We made fairness computed and visible: distribution of weekends, nights, and undesirable shifts per person over a rolling window, shown to the scheduler as they build and to staff as published. A roster that is coverage-optimal but distribution-skewed now reads as a warning state rather than a success. The swap marketplace lets staff resolve their own conflicts within the fatigue rules, without a manager in the loop, which removed both a bottleneck and a source of perceived favouritism. Turnover moved because the system stopped producing outcomes people experienced as arbitrary.
THE INTERFACE CRAFT
The interaction, rendered as a working product surface.
The specimen below is code-native and uses the study's own design logic. The client interface remains protected.
ENTERPRISE INTERNAL
SHIFT / SATURDAY
08:00Coverage 92%
16:002 open
00:00Covered
DESIGN DECISIONS
Positions we would defend.
Each decision names the principle and the product consequence, not a stylistic preference.
Fairness is a variable
A scheduling system with no representation of fairness will produce unfair schedules and be unable to explain why.
Remove the manager from routine swaps
Manager approval on every swap creates a bottleneck and a perception of favouritism. Automate the legal ones.
Publish the distribution
Staff accept an unfavourable roster far more readily when they can see it is not unfavourable every time.
PRODUCT LEADER READOUT
What transfers, and what should remain specific to this product.
A case study is useful when its operating principle travels without turning the original interface into a template.
Read the operating condition
For Workforce management · healthcare system, the transferable lesson is not a copied screen. It is the condition the interface had to make legible: Weekend, night and holiday distribution per person over a rolling window, visible while building and after publishing. Skew is a warning state. Rebuild that visibility for your own roles, risk, terminology, and operating cadence.
Protect the design rule
A scheduling system with no representation of fairness will produce unfair schedules and be unable to explain why. Keep that rule in the acceptance criteria, component states, and production QA record so later visual cleanup cannot erase why the interaction exists.
Measure behaviour after ship
The evidence record is -22% for voluntary turnover, first two quarters. Recreate the baseline and outcome window before rollout, segment the result by role and context, and state clearly what the measure cannot prove.
RESEARCH RECORD
The work behind the interface.
These artefacts connect the final interaction back to the evidence and product model that produced it.
Night-shift ethnography
Observed three full night shifts and interviewed 24 nurses about how they experienced the roster, not how they used the tool.
Exit interview analysis
Coded 90 exit interviews; scheduling appeared in 34% as a contributing factor, second only to compensation.
Fairness metric design
Built the distribution metric with nursing leadership; four definitions tested before one was accepted as legitimate.
Swap legality engine
Encoded the fatigue and union rules with HR and legal so automated swaps could not produce a violation.
“It stopped showing up in exit interviews. I have run this unit for eleven years and that has never happened.”
Director of Nursing, hospital system · under NDA
NEXT STUDY · 17 / 30 · ENTERPRISE INTERNAL