System first, then the product
Rebuilding EventBookings solo, on a design system.
EventBookings is one of Australia's biggest ticketing platforms. When I joined the redesign I was the only designer, working with a product manager, engineering and QA. There were dozens of screens to fix, and one of me.
So I started with the system, not the screens.
Why the system came first
As the only designer, consistency couldn't depend on me remembering every screen. If I redrew pages one by one, they'd drift. If I built the pieces first, new screens could be assembled rather than drawn, and engineering could build from the same pieces.
The system, EBDS, came in two parts:
- Foundations: a colour scale, a Poppins type ramp, spacing, elevation, icons, illustration and avatars.
- Components: 24 of them, from buttons and text fields to tables, modals and side navigation, with every state designed: normal, hover, active, focused, selected and disabled.
Then the product, where people actually failed
Analytics in Mixpanel and Google Analytics showed where people dropped off. Organizer interviews and support tickets explained why. Three problems stood out: a slow checkout, a confusing event setup, and a dated interface.
- Checkout: five-plus steps became three (pick tickets, check out, pay), with the real price on the first screen. Checkout time fell from 2m 40s to 1m 05s, and cart abandonment dropped 28%.
- Event creation: 8 to 10 confusing steps became five guided sections with smart defaults and drafts that save themselves. Creation time dropped 40%, monthly new events rose 31%, and support tickets fell 45%.
- One workspace: in phase two, fifteen organizer tools (tickets, sessions, seating, insights) moved into one layout with the same header, navigation and list patterns. Learn one screen and you know them all.
A system isn't overhead for a solo designer. It's how one person moves like a team.
What I'd tell another solo designer
- Design every state up front. Disabled, loading and error states are where systems fall apart later.
- Fix the system, not the screen. When a page looked wrong, I fixed the component, and every screen using it improved at once.
- Measure after launch. The system made the redesign possible; the numbers proved it worked. Over the project, revenue grew 611% and traffic 242%.