UFO Design System
As design lead and system owner, I identified the need myself, with no roadmap ask, and built the component and motion library that let Amazon's installment offer scale to 33 countries without being rebuilt in every market.
Short on time? Pick your read.
Full case study — process, tradeoffs, and craft.
A problem nobody assigned me
There was no ticket for this, no roadmap line, no directive from leadership. As Priceblock and Compare Plans launched market by market, I kept seeing the same pattern repeat: each new region rebuilt its own comparison card from scratch, because the old Twister+/AUI approach handed every market a blank page instead of a shared foundation. Nobody asked me to fix that — I identified the risk myself and made the case that it was worth solving before it became a much bigger cleanup problem.
So I proposed and built the UFO Design System on my own initiative — a reusable, AUI-independent library of components, states, and motion that now powers Priceblock and the Compare Plans table across every marketplace.
Design once, scale everywhere
I built this in parallel with my assigned feature work, without dedicated headcount or timeline for the system itself. Documented specs and Figma files let six design partners across North America, Europe, Asia, South America, and MENA onboard fast and adapt the library to their own market — I stayed the bar raiser throughout, reviewing every market's implementation before engineering work began there.
The guiding principle: customers should never see the seams. Even when markets run on divergent tech under the hood, the way offers are presented, ranked, and selected stays consistent — so the experience feels like one product everywhere.
I defined the anatomy, interaction states, and variant rules for the comparison card, then codified the motion — emphasized easing for container transforms, spring curves for physical feedback — as reusable tokens. A junior designer helped populate variants in the library from that anatomy; I audited and edited every one for quality before it shipped. Two engineers partnered with me to confirm what I was proposing was actually buildable.
Rollout was iterative by necessity: build, launch, audit the next market's needs, update the library, repeat. I ran worldwide design reviews myself to keep quality consistent as coverage grew.
The comparison table's growth was capped by a real engineering constraint: the desktop side sheet it lived in didn't support horizontal scrolling, so it couldn't render more than 3 offers without breaking. Working with engineering, I proposed rebuilding the container as a Modal component instead — which not only solved the scrolling limitation but let the table show 5 offers at once with room to add more providers over time, turning a technical fix into a UI improvement.
Charlie’s exceptional ability lies in his visionary approach to scalability, evident in his valuable contributions to the UFO initiatives. His collaborative thinking and emphasis on scaling ideas make him an invaluable asset, especially in the dynamic environment of Amazon.
Same system, three audiences
Tap a piece of the system, then switch lenses. The words change more than the facts do — that contrast is the point.
I just want to see my payment options side by side, not hunt for a plan that might be hiding in a different tab.
One comparison-card anatomy — structure, interaction states, and variant rules — documented as a single source of truth. AUI-independent, so it doesn’t inherit the constraints that capped the old Twister+ approach.
Powers Priceblock and the Compare Plans table across every marketplace — 5 documented card states and variants, one definition instead of a market-by-market rebuild.
When I tap a plan, it should feel like it’s actually responding to me — not just popping into place.
Material 3 motion codified as reusable tokens — emphasized easing for container transforms, spring curves for physical feedback — so animation stays consistent across surfaces instead of being redefined per screen.
One motion vocabulary, reused everywhere the card appears — no market or surface ships a slightly-different feeling version of the same interaction.
Not my problem how many countries this launches in — I just want the same experience wherever I’m shopping from.
Documented specs and Figma files let 6 design partners across regions onboard without hand-holding. I bar-raise every market's implementation before engineering starts, running worldwide design reviews myself — no dedicated headcount behind the system itself, even as the org using it has grown.
33 countries on one shared system, WW governance self-run from day one — scale without reinventing the card in every market.
Measurable outcomes
Framework 2.0 shipped August 2024 — a follow-up survey is tracking whether partner satisfaction has climbed past that initial 85%.
The system is the product
Turning the installments experience into a documented, motion-aware component library is what let it scale to 33 countries across six regions — and a projected $900M expansion — without re-solving the same problems each time.
A design system isn't a deliverable — it's leverage. The best ones let a whole org move faster while raising the bar for the customer.
If I were starting over, I'd have designed for international scale from day one instead of as an afterthought — the initial anatomy and motion tokens assumed a single-market rollout, and scaling to 33 countries meant reopening decisions that should have shipped with regional variation built in.
When I handed the junior designer the anatomy and interaction rules to start populating variants, I didn't give clear enough guidance on how the variances and interaction states should actually be structured inside the library itself. Several days into the work, we found that my instructions hadn't translated the way I'd assumed — a chunk of what had been built didn't match the variant logic, and it had to be redone. That was lost time I could have prevented with a clearer spec up front, not something I caught early enough to redirect before the cost was already sunk.
Leverage other teams could run without me
A design system is only leverage if it changes what other teams can do on their own. Running WW design governance meant reviewing every market's implementation before engineering started there — and once Framework 2.0 shipped, a market that previously needed weeks of bespoke component work could onboard roughly 2 weeks faster, with 85% of surveyed partner teams reporting a positive experience with the library.
I also ran the cadence that kept the system honest as it grew: a 3-day cross-team workshop with Research and the 3P team to align on day-to-day friction and design goals, and regular working sessions to source precedent and stress-test new discoverability ideas before they reached engineering.
I proposed testing installments placement directly next to price on the Detail Page, to fix a real discoverability gap. The team that owns that surface pushed back hard, citing valid concerns about layout scalability across every product listing. Rather than drop the idea or force it through, I offered guardrails — limiting the test to specific product categories or a single platform — to de-risk the experiment while still generating a real read on customer behavior.
They weren’t willing to engage on guardrails, which put two planned discoverability experiments at risk for the rest of the year. Rather than let it stall quietly, I partnered with Product to escalate through the proper channel — a joint escalation document presented to senior leadership — instead of going around the other team or dropping the idea. That's the outcome I'd point to, not the win: knowing when a disagreement is worth escalating formally, and doing it in partnership with the team you disagree with rather than around them.
Charlie, what a year you’ve had! Your accomplishments demonstrate performance that meets Amazon’s high performance bar for your role and level. Starting with the most impressive stretch goal, you led the full E2E redesign of the installment discovery CX in Q4 2025, collaborating across multiple partner teams to influence and address customer pain points that simplify installment selection. This stretch goal was delivered all while keeping up with your daily deliveries, where you led the launch of UFO in cart across US and BR, and gathered alignment with the APEX team to launch two high-impact experiments on detail page that amplified monthly price through a font size experiment and a one-click selection CX in JP. Both significantly increased Amazon OPS.