One of the issues Mercari faced was that they had many inconsistencies across components in their existing application. Before a rebrand could be carried out, it was essential for me to first homogenise these components under a single source of truth (design system).
As Mercari was in fail fast learn fast cycle, we iterated very quickly to determine what works and what didn’t work.
As it was essential for me to first understand all the standard components across the app before starting a design system, I began mapping out what Mercari’s current app had looked like at the time. After that, I began to catalogue standard text and layer styles which were shared and began to build out components, breaking them down by atoms, molecules, and organisms.
Native iOS & Android
I worked closely with data scientists and analysts to determine how design and UX decisions can impact GMV (Gross merchandise volume) and 7-day retention rates, which were KPIs for the business.
We had stringent KPIs to meet at Mercari, mostly centred around 7 Day Retention and GMV (Gross merchandise volume). Every design decision made was user tested before being rolled out to a limited alpha testing audience, and then A / B tested on the production app. We would monitor GMV from both A & B groups and attempted to determine if a design change resulted in an increase or decrease in GMV between the two groups.
More often than not, we would test up to 6 variations of a single component in an A / B test. We would rapidly iterate our designs based on what worked based on the data and what didn’t, we learned and moved forward.
To further improve the Mercari experience and provide a stickier and more engaging experience, I came up with various animations that were tested in the app. The animations were created in after effects and after that, exported as JSON files for Lottie integration on iOS and AVD for Android.
Sometimes purely basing the design off data can make you lose sight of the bigger picture. It was important for Mercari to have a Northstar vision in place so that we can continue to provide a marketplace that was engaging for millennials and gen-z audiences.
I explored many concepts which were rapidly designed and prototyped within 1-week time blocks using the Google Design Sprint methodology; these concepts were then exhaustively tested in different regions before we would pick the best from the crop to implement.