Its cost summary is used always or often by 81% of surveyed users: a redesign of the screen every manager on the floor lives in.
Five years of features, added one on top of another, had left it capable of almost everything, and that was the problem. My job wasn't to add anything; it was to make all of it legible again. Years later, a survey of 397 users found 81% use its cost summary always or often.
At a glanceA header with one job, a persistent sidebar, and stacked modals turned into tabs.
Read → 04 The shift that never savedTesting showed every participant thought their quick shift was saved. None were, and that set the brief for what came next.
Read → 05 Scheduling against a live cost targetA red and green budget signal that, years later, is the number users steer by.
Read →The Schedule Editor is where a manager builds the weekly roster: who works which shift, in which role, against a budget. It's one of the platform's most used screens, and five years of features added one on top of another had left it dense, slow and hard to read. Opening it for the first time, you couldn't tell what half of it did.
My real move wasn't a nicer toolbar; it was redistribution. Instead of cramming every control into the header, I gave the header a single job (pick a location, move between weeks) and pushed everything else down into the places it's actually used. Managers already knew the week grid, so I kept it and rebuilt around it: a persistent sidebar (calendar, week/day toggle, filters and preferences that never move); the old stacked modals turned into tabs: Editor, Shift Warnings, Requests, Notes, Timesheet, History, each with a count so you can see what needs attention; and the three decisions that matter, Approve, Payroll, Publish, pulled out where you can't miss them.
Managers don't all picture their floor the same way: some think in people, some in sections, some in job or department. So I let them regroup the same schedule into Employee, Job, Section or Department views, with a label that always says which one you're in. Colour does the heavy lifting: every section owns a colour that stays consistent across the grid, the filters and the open shift chips, so the eye can track a team without reading a word.
Some problems only show up when someone else is driving. I ran unmoderated usability testing on the prototype through Userlytics, with managers and executives from its participant panel, each asked to create a quick shift, an advanced shift and an absence while thinking aloud. The advanced shift worked for everyone. The quick shift worked for no one: every participant typed it in, clicked away and believed they were done, but nothing had saved. They also couldn't tell which of the two modes they were in, because the plus and arrow icons gave no hint before you clicked.
The pattern was unanimous, and it gave the next iteration a clear brief:
A schedule isn't just who works when: it's a labour bill measured against forecast sales. What they also needed was a summary bar that sits under the week and ties it together per day: forecast and actual sales, forecast labour, the salary and hourly breakdown, and a budget percentage that turns red or green as you go over or under. Compliance lives right in the grid too: an under 18 flag, overlap and overtime warnings, and a projected vs scheduled hours model once breaks are applied. That means a manager is making cost and legal calls while they schedule, not after.
Years after launch, a survey of the scheduling report (397 completed responses) put the bar to the test: 81% use it always or often, two thirds every single time they build or adjust a schedule, and labour percentage with its red and green cue came out as the one number almost everyone steers by.
"The first thing I check is whether I am in the green and by how much."
Respondent, scheduling report user survey
A finished week travels: Approve → Payroll → Publish, across a manager, an approver and payroll. I designed that journey as a series of clear states rather than silent saves, and made the messaging carry weight. A "sent" banner you can dismiss; a rejected banner you can't, that stays put until the conflict is resolved and links to exactly what needs fixing. And I split publishing into plain choices: updates only, all shifts, or open shifts, so a manager always knows exactly who gets notified.
The hard part of a redesign like this is restraint. The tool couldn't lose a single thing it could do, so almost all of the work was rehoming: deciding what deserves to be visible, what belongs one level down, and what should follow the manager into a flow rather than sit on the surface. Watching a five year tangle become legible again, without giving up any of its power, is the part I'm proudest of.
I led and originated the redesign; the platform's new component library was later applied on top by a colleague, with minor adjustments. Delivered with a cross functional team across product, engineering and QA. Focuses on the design problem, process and decisions; outcome figures come from a later user survey of the scheduling report (397 completed responses) and unmoderated usability testing on Userlytics.