Sign-Up Flow Overhaul
Carousell is a C2C marketplace operating across Southeast Asia, Taiwan, and Hong Kong. By 2022, the product was mature, but the data told a different story at the top of the funnel.
Only 20% of new users were completing the sign-up flow. That meant 8 out of 10 people who tried to join never made it through. As the sole designer on this project, I led the end-to-end redesign of the sign-up experience, from research through to final delivery, with the goal of removing friction early enough to give users a reason to stay.
Given the resource constraints we were working with, we launched in phases, tackling the highest-impact changes first.
I started by pulling six months of funnel data to map where users were dropping off. The pattern was clear: the biggest drop-offs weren't on the account creation steps — entering an email or setting a password. They were on the personalization steps: selecting interest categories, following recommended sellers, and setting up a feed before users had any context for what the app actually offered. We were asking users to make decisions about a product they hadn't experienced yet.
To understand the "why" behind the numbers, I conducted interviews with 8 users across three groups: people who had never used Carousell, casual users, and more active sellers. The sessions revealed a consistent pattern, users felt the app was pushing a seller identity on them before they'd decided how they wanted to use it. As one user put it: "It's a lot to do with selling phones, I'm getting a feeling I need to sell something — feels like a seller-oriented app." Several others tried to skip through the personalization steps as fast as possible just to get to the product.
The flow asked too much, too soon. Users were required to select interests, follow sellers, and configure a feed before they'd seen what Carousell had to offer. The steps weren't wrong — the placement was.
The onboarding created the wrong first impression. Seller-focused choices early in the flow made buyers feel the app wasn't built for them. A single undifferentiated flow was failing three distinct user types: buyers, casual sellers, and merchants.
Several screens lacked clarity. Outdated UI patterns and ambiguous instructions left users making decisions without enough context to make them confidently.
To move from problems to solutions, I ran a workshop with the PM and researcher. Working from the three personas, we framed the challenge as a single HMW:
💡
"How might we design a concise and quick sign-up flow that only asks necessary questions, allowing users to experience the product first."
The strongest idea to emerge was a personalized entry point — letting users identify as a buyer, casual seller, or merchant at the start, then tailoring the flow to match. It directly addressed the wrong-first-impression problem and would have made the experience meaningfully different for each user type.
New flow: Personalized entry point
Rather than delay the entire release, I worked with the PM to phase the work. Phase 1 focused on the highest-impact, lowest-complexity changes: removing confusing screens, simplifying personalization, and updating UI components to the new design system.
The personalized entry point — which required significant backend work — moved to Phase 2.It wasn't the ideal solution. But a meaningfully better experience now was more valuable than a perfect one later.
Ideation
With the scope defined, I focused the final design on three changes that directly addressed the problems we'd identified. Simplified the flow from 6 steps to 3. The removed steps were the personalization-heavy ones — selecting interest categories, following recommended sellers, and configuring a feed. These weren't deleted permanently; they moved to post-onboarding, where users would have enough context to make those choices meaningfully. What remained: account creation, minimal profile setup, and a direct path into the product.
We simplified personalization to just gender and age. Instead of asking users to manually curate their experience upfront, recommendations became automatically tailored based on basic inputs. Less friction, same personalization outcome.
Updated UI to the new design system. Outdated components were replaced, ambiguous instructions rewritten, and visual hierarchy clarified across every screen.
For the web version, I extended the same flow while adapting it for the platform's constraints. Unlike mobile, web doesn't have native components for things like date pickers, dropdowns, or permission prompts, each had to be rebuilt using web-native patterns while maintaining visual consistency with the app.
The goal was for a user moving between platforms to feel like they were in the same product, even if the underlying components were different.
Phase 1 shipped on schedule. The results were meaningful:
Two things I'd carry into the next project:
Constraints are a design input, not just a limitation. The timeline pressure that forced us into phases ended up being clarifying, it made us ask which changes actually mattered most. The personalized entry point was the "right" solution, but Phase 1 proved we didn't need it to make a significant impact. Sometimes shipping the 80% solution teaches you more than waiting for the 100% one.
Phased releases need a plan, not just a promise. We shipped Phase 1 with a clear intent to follow up, but "Phase 2" without a concrete timeline is just a good intention. Next time, I'd push for Phase 2 to be scoped and scheduled before Phase 1 ships, not after.
Next: Design System Revamp →
