Roblox

When Play Feels
Socially Risky

Before kids decide what to play, they decide whether they feel safe enough to participate.

Role Product Design Manager
Scope Avatar · Discovery · Social · Onboarding
Focus Pre-IPO Growth · Research Strategy · Social Product
Roblox social scene — players in a shared environment

I was leading product design at Roblox when the team was asked to improve game discovery. Research we ran revealed the real problem was social confidence — not navigation. That reframe changed what we prioritized, what we built, and how the product team was organized.

The problem wasn’t discovery.

It was social confidence.

Roblox asked how to help kids find more games. The assumption was navigational: better recommendations, clearer categories, smarter search.

The data told a different story. Kids were already finding games. The hesitation happened right before they joined them.

Entering a game on Roblox wasn't purely a gameplay decision — it was a social one. The moment you stepped in, other players could see how your avatar looked, how you communicated, whether you seemed like someone who belonged there.

For younger players especially, that moment carried real emotional weight. Internally, Avatar, Discovery, Onboarding, and Social were treated as separate product areas. But kids experienced them as a single continuous system — one that either prepared them to participate socially, or left them exposed.

The question underneath discovery

“What helps kids feel safe enough to participate?”

What the company assumed
What research revealed
Kids can’t find games
→
Kids hesitate before joining them
Discovery problem
→
Social confidence problem
Better recommendations
→
Confidence to participate
The moment of game entry
BROWSE Scrolling home CONSIDER Views game page HESITATE ← the real blocker "Will they judge me?" "Do I look like a noob?" "Will I fit in?" ENTERS feels ready LEAVES fear wins Discovery led to a moment of social hesitation.

What kept surfacing

Choosing a game also meant choosing how to be seen.

“I feel unwelcome whenever somebody says my character isn’t cute, or that I don’t have enough money to look good.”

Research participant · Roblox player, age 10

That quote reframed the problem. A default avatar wasn't just cosmetic — it could mark a player as inexperienced or exposed before they had said a word to anyone.

Many players joined games because their friends were there. If their friends weren't online, they often didn't want to play at all. The game itself mattered less than the social context surrounding it.

Players weren't simply browsing content. They were deciding whether they felt ready to be seen.

Where the participation flow breaks
Identity Expression Confidence Am I ready? Social Readiness Exploration What to enter? Participa- tion

Many kids disengaged socially before they stopped playing Roblox.

The leadership moment

Making the case to slow down.

The instinct across the organization was to move quickly. Discovery metrics existed — engagement rates, session starts, funnel drop-off. The pressure was to optimize what was already measurable.

I pushed back.

Metrics show behavior, not motivation.

Optimizing recommendations without understanding why kids were hesitating would mean shipping the wrong solution — one that looked like progress and wasn't. That argument, made to PMs and leadership, created the space to step back and study how kids actually decide whether to participate.

That finding gave me the grounds to redirect the work entirely.

How product direction shifted.

That changed what we prioritized.

Once confidence became the real problem, the work shifted from optimizing discovery toward helping players feel more prepared to participate socially.

1. Redesign the avatar experience
What had been an inventory-style catalog became an immersive, identity-first experience. The design direction shifted the catalog, the editor, and the customization flow away from transaction and toward exploration — treating self-expression as something players could experiment with, not just select from a grid. The goal wasn't cosmetic personalization. It was helping players feel ownership over how they showed up in the world.
Shipped
2. Reconsider what onboarding was asking of players
Early onboarding required players to choose an avatar before they had any real sense of what Roblox was or who they wanted to be there. That decision — made at the most unfamiliar moment — created friction before players had the context to make it well. The better design bet was clear: rather than asking players to declare themselves upfront, let them discover who they wanted to be through play.
3. Communication beyond chat
Roblox had strong moderation systems designed to protect younger players. But heavy reliance on text created friction — particularly for players who weren't yet comfortable expressing themselves that way. That led to explorations around reactions, gestures, and lightweight visual signaling: ways to participate socially without requiring language.
Before Original catalog — inventory-style, light UI, item-first
Original Roblox avatar catalog — inventory grid with category sidebar
After — shipped Redesigned catalog and editor — immersive, identity-first
Redesigned Roblox avatar catalog and editor — dark UI, character-centric, identity-focused
From inventory management to identity exploration. The catalog and editor were redesigned as a single expressive system — treating how a player looks as a social statement, not a shopping list.
Onboarding exploration From character selection to identity customization
Old and new Roblox onboarding — character lineup vs. avatar customization
Before: choose a character before entry. After: begin shaping one — with context, through play.
Design vision Social-aware discovery — peek view for the intermediate funnel
Design vision for social-aware discovery funnel with peek view states for new and returning players
A social-aware discovery system — peek views, friend context, and session signals that let players evaluate social fit before committing to entry. Different states for new players and returning players reflect the different confidence and context they bring.

What the work changed

The reframing had structural consequences.

The most visible shift was in the product. But the more consequential change was organizational.

The argument I made to leadership — that social confidence, not content discovery, was the real blocker — led to a structural change. The Social team, previously focused on safety and moderation, expanded its mandate to cover participation and social readiness. Safety asks: what do we protect players from? Participation asks: what helps players feel ready to show up? Those are different problems that require different design investment — and the org needed to treat them that way.

I also repositioned how the team used research — from validating solutions to problems already assumed, to questioning whether the right problems were being considered at all. That shift was as important as any individual product decision.

What the work made visible

What first looked like a discovery problem was really a confidence problem.

For a 10-year-old on Roblox, the barrier to entry isn't usually that they can't find something to play. It's that they're not sure they'll be accepted when they get there. That's a harder problem to instrument. It doesn't show up in session-start rates or recommendation CTR. It shows up in the kids who stopped opening the app.

Before people explore, socialize, or contribute, they first need to feel ready to show up. Products that miss that miss the actual barrier.

That's the lens I bring to any product with a participation problem — design for the moment before someone decides to show up, not just for what happens after they do.

Kids didn’t need better recommendations first.
They needed confidence to participate.

Carina Ngai Product Design Manager

Stella Chen Product Designer

Sean Donahue Principal, RCD/LA Core Faculty, Graduate Media Design Practices (MDP), ArtCenter College of Design

Kenny Jou Lead Product Manager, Game Discovery Vertical

Anant Agrawal Associate Product Manager