Tencent · QQ · UX Design Intern
Designing for 600 million people
on a 20-year-old product.
Where habits were the hardest thing to redesign.
What I worked on
Two of QQ's most-touched surfaces: the share panel and the message hub. Both looked like UI cleanup on the surface. They turned out to be the same harder question, asked twice — when a product has been used for 20 years, how do you change it without making 600 million people feel lost?
Shipped to production
iOS, Android, and tablet. Gray-release validated. ~80% of high-frequency users preferred the simplified share flow. Message hub shipped with the simplified primary view and no per-category badges.
Setting
When a product evolves, whose habits get to break?
QQ has 600 million monthly active users and more than 20 years of history. When something has been used that long, the interface isn't really the interface anymore. It's muscle memory. Hundreds of millions of people, one share at a time, one notification at a time.
I worked on two surfaces. Both projects looked like UI cleanup. They turned out to be the same harder question, asked twice.
Chapter one · The share panel
The share panel
QQ had just rolled out a unified share panel. Cleaner visuals, a more cohesive design language, and a structure that could grow with future features. From a systems perspective, it was a thoughtful upgrade.
Within weeks, users were unhappy.
“I used to share without thinking.”
“Now I have to stop and choose.”
“It feels slower even though it looks better.”
The new panel had added exactly one step to the share flow. Two taps became three. Functionally tiny. Emotionally enormous.
Pull up the sharing panel
Figures here are directional and presented under Tencent's NDA.
The complaints clustered in one group. The people for whom sharing was reflexive, not deliberate.
The real question wasn't “make sharing faster.” It was, how do we make it feel fast again for the people who share constantly, without throwing away a system the team had just built?
Three directions
Bring back the two-tap flow exactly as it was.
Fast for users. Slow for the platform's long-term shape.
Better visual hierarchy, clearer affordances.
Same structural friction. A prettier version of the same problem.
Keep the unified panel where flexibility matters. Restore the fast path where speed matters.
Shipped. Structural investment intact. Muscle memory restored.
We shipped the third one. A small, lightweight component for the most common share moments, sitting inside the unified panel as a quick path. The new system stayed intact. The muscle memory came back.
This change in the design system changed user behavior.
Before
After
Gray-release validated. Roughly 80% of high-frequency users preferred the simplified flow.
After this solution launched, sharing felt noticeably smoother. We received a lot of positive feedback from users, especially from high-frequency users. They strongly felt the reduction in steps.
Chapter two · The message hub
The message hub
The second project started with the same kind of complaint and ended at the opposite kind of answer.
QQ's message hub mixed every interaction type into one stream. Likes. Comments. Shares. Ratings. As volume grew, people felt buried.
“Likes drown out everything else.”
“I miss important replies.”
“I want categories.”
The instinct was obvious. Separate the stream into category tabs. The PM wanted it. The most vocal users wanted it. On paper, it solved the problem.
Then we segmented the user base, and the picture changed.
The most vocal feedback came from the 20%. The structural decision would affect the 80%.
Loudest feedback isn't the largest user group. And complexity has a cost that doesn't show up in feature requests.
I argued, with both behavioral data and interview transcripts, that we should design for the silent majority. Highly active users are already deeply engaged. They aren't leaving. Regular users are quieter, and their experience is what drives retention.
What shipped
Fast scanning, easy clearing. Serves the 80%.
Power users keep their separation when they want it.
Tempting on paper. But it would have meant backend complexity for the team and more cognitive load for everyone. This was the hardest call to defend. It's also the one I'm proudest of.
The simpler thing to do is add the thing. The harder thing is to defend the empty space.
Takeaway
What I took away
Two projects, same lesson at the end.
When you're working on a product that has been around for 20 years, most of the design has already been done. The users did it. In their hands, over time. The job isn't to introduce something better. The job is to evolve the product without making people feel lost.
I went in thinking the work would be visual. I left understanding it was archaeological.
The phrase that kept coming back in feedback was “it feels smooth again.” What I noticed most was the verb. Feel, not work. The interface had never been broken. The feeling had. Good redesign is not about “loud” change. It is about growing without losing the familiarity and clarity people value.
Want the full story?
The unabridged walkthrough has the exact metrics, full process, and unredacted screens.
Happy to share it on request.
Get in touch →This case study is published with Tencent's NDA in mind. Specifics have been generalized and some screens have been simplified.