Carousell is a marketplace for buying and selling across Southeast Asia, Taiwan and Hong Kong. By 2022, only 60% of new users finished signing up. Four in ten left before they even started. I led the redesign of the sign-up flow to remove friction and give people a reason to stay.
Six months of funnel data showed that the biggest drop was not account creation. It was the personalization steps after it: picking interests, following sellers and setting up a feed. We were asking people to make choices about a product they had not used yet.
I interviewed 8 users across three groups: newcomers, casual users and active sellers. The same thing came up again and again. The app pushed a seller identity before people had decided how to use it. One person said, “I’m getting a feeling I need to sell something.”
The best marketplace apps I looked at did the opposite: get people to their first useful moment fast, and ask about preferences later.
To decide what to fix, I ran a workshop with our PM and researcher. We landed on one question: how might we tailor sign-up to what each person came to do?
The strongest idea was a personalized start. People pick buyer, casual seller or merchant, and each gets its own flow. It fixed the wrong first impression directly. But it needed a lot of backend work, and the timeline was already at risk.
So I made a call with the PM: a clearly better sign-up now, instead of a perfect one later. We split the project. Phase 1 took the changes with the most impact for the least effort, and the personalized start moved to phase 2.
For web, I rebuilt native parts like date pickers, dropdowns and permission prompts with web patterns, so both versions feel the same.
Phase 1 shipped on schedule.
The deadline that forced us to split the work also showed what mattered most. Phase 1 got most of the impact without the personalized start.
But phase 2 never shipped. A phase 2 with no scope and no date is just a good intention. Next time, I’d plan it before phase 1 ships.
I’d also measure something else. Finishing sign-up isn’t the real goal; using the product is. I’d define the first-week action that predicts whether someone stays, and design sign-up to lead there.
