Apple · 2020
Simplifying global artist administration
Expert teams spent their day searching a tool that had grown one screen at a time.
Interface recreated for portfolio presentation to protect confidential product information.
My role
Product Designer
Impact
- 25% faster search
- 40% fewer clicks
- 27% lower operational costs
Scope
- Workflow design
- Search experience
- Information architecture
- Design system foundations
Timeline
2020
The challenge
An administrative tool grown by accretion: every team had added a screen and nobody had removed one.
The opportunity
Rebuild around the small number of tasks that carried almost all of the volume.
What research changed
A few tasks carry the day
A handful of workflows accounted for most of the actual use.
Search had become navigation
People searched because the information architecture had stopped being legible.
Bulk is the norm
Very little work happened one record at a time.
How might we
How might we let an expert finish a routine task without ever leaving the results view?
Directions explored
- 01
Restructured IA
Cleaner, and still required navigating away in order to act.
- 02
Modal-driven editing
Fast for one record, punishing at scale.
- 03
Search as workspace
Filter, select and act inline.
Decisions
Decision 01
Design for density and keyboard flow.
- Optimise for new users
- Optimise for daily expert users
Options considered
- What changed my mind
- The population was small, stable and highly practised — onboarding friction cost far less than daily friction.
- Why
- Minutes saved per task, many times a day, was where the value was.
- Tradeoff
- A steeper first hour for anyone new to the tool.
Decision 02
Reversible actions instead of confirmations.
- Confirmation dialogs
- Undo
Options considered
- What changed my mind
- Confirmations were being dismissed reflexively, so they protected nobody while slowing everybody.
- Why
- Speed and safety only coexist when mistakes are cheap to correct.
- Tradeoff
- Undo required state handling that a dialog would not have needed.
Constraints we designed around
Legacy-system dependency
Existing records and permissions models could not be redesigned, so the new workflow had to sit on top of them.
Where it landed
Persistent search workspace
Query, results and actions in one uninterrupted surface.
Inline bulk actions
Multi-select with a single reversible commit.
Shared system foundations
Tokens and components reused by neighbouring tools.
Outcome
25%
Faster search
40%
Fewer clicks
27%
Lower operational costs
What I took from it
Enterprise craft is mostly subtraction. The measurable wins came from the screens we removed.
Next project
Meijer
AI-Powered Associate Experience