Keeping 6 designers consistent across 3 platforms in 4 weeks

Refresh the entire visual language across three platforms in four weeks, without changing a single user flow, and without the final brand in hand.

Loco was redesigning its viewer experience across the app, web, and mobile web. For its audience, the app is the product, so a visual overhaul of that surface carried real risk.

The deadline was set by an external event, not by us. The refreshed product was going to be shown at iGB L!VE in London, an industry event at ExCeL that draws around 15,000 attendees. That was the version global partners and press would see, so the date could not move by a single day.

Working backwards from that date produced four constraints that shaped every decision after it.

Product Designer on the viewer experience, owning four of the highest-traffic surfaces alongside the shared component work the rest of the team built on.

I was responsible for designing several high visibility experiences, including the Homepage, Livestream, Fullscreen Livestream, and Leaderboard.

Rather than recreating reference screens one by one, I focused on building reusable components that could be shared across the team. My goal was to make each screen easy to maintain while keeping the experience consistent across different platforms.

Since several designers were working in parallel, regular conversations became just as important as the components themselves. We frequently aligned on visual decisions to reduce differences before they spread across the product.

Rather than treat the constraints as background, I used them to set scope. Each one had a design response, and naming those responses early is what kept six people building the same product.

Engineering asked for tokens to simplify light and dark theming. The bigger use case was that tokens would let the whole team build the product against a brand that had not been delivered yet.

By referencing semantic roles instead of fixed hex values, the missing identity stopped being a blocker and became a variable. Everyone could build on placeholders, and the real brand could be swapped in later, everywhere, at once.

Every component I built referenced tokens rather than colours. At the time it felt like a technical improvement. It turned out to be the decision that saved the launch.

The library had to serve three platforms and five other designers at once, which meant components needed to hold up in contexts I would never see myself. Building for that from the start cost more in week one and paid back every week after.

Around 85% of screens shipped entirely from shared components, with no bespoke mockups needed.

With four weeks, it was not practical to mock every screen before development began, so we let the system carry the specification instead. Clear styling guidelines acted as the single source of truth, which let engineers build most screens without waiting on detailed mock ups.

That only works if someone answers the gaps quickly. My teammate and I stayed embedded through implementation, tracking Slack and staying available in the war room, usually resolving component questions within the hour. Some needed a new screen, most only needed a decision on how an existing component should behave.

What I took from it was a sharper read on what engineers need under deadline. Their questions were rarely about the design itself. They were about reducing uncertainty so they could move forward with confidence. A well structured system answers most of those before they are asked, and being present answers the rest fast enough that builds never stall on design.

Three weeks in, the agency delivered the final brand: new colours, gradients, iconography, with one week left before launch.

Because every component referenced semantic tokens, we applied the new identity across the entire product in a single evening. The manual alternative was estimated at two to three weeks of screen-by-screen updates, which would have pushed us past the iGB L!VE date. A quiet decision made in week one saved the launch in week three.

The success criterion was continuity: change how the product looks without changing how it behaves. Thirty days after launch, every headline engagement metric sat within a quarter of a percent of its pre-launch value, and around 85% of screens shipped entirely from shared components with no bespoke mockups needed.

For a redesign whose explicit goal was continuity, holding 160k active users steady through a full rebrand is the result: a new identity introduced without costing the business a point of engagement, and no spike in UI-related support tickets through the transition.

Loco was pushing into new international markets around the same time, and the redesigned experience became the front door for that expansion. In Vietnam, traffic climbed from a near-flat baseline to a steady state several times higher, and the quality of those sessions held: a 14.2% bounce rate, an average session of 8m 6s, and 2.46 sessions per user, meaning people were returning rather than passing through.

• The system outlived the launch. The token architecture and component library remain the foundation for new feature work today.
• Theme support shipped essentially for free in a later release, the tokens' original purpose, delivered as a byproduct.
• The post-launch audit found inconsistencies on under 10% of screens, resolved in one cleanup sprint and folded back into the guidelines.

Creating reusable components is only part of the job. The bigger challenge is helping people use them consistently. Even with a well structured system, six designers under deadline pressure will make six slightly different calls, however good the library is.

Looking back, I would introduce lightweight checkpoints throughout the project instead of reviewing everything after the work was complete. A fifteen minute shared component review before any flow was marked ready would have caught most inconsistencies at the source, without slowing the team down.