Nobody could find the market they wanted.
Thousands of markets, and users kept asking for more. I redesigned how people search and browse Futuur, one design improvement at a time, measuring each change on its own.
- Role
- Sole product designer — research, interaction and visual design
- Team
- CEO, engineers, operations
- Timeline
- 2025 – 2026, four releases
| Pages per user after the navigation redesign | +84% |
|---|---|
| Engaged time per session | +49% |
| Bounce rate | −12 pts |
"I wish you had more markets about what I like."
We heard this in almost every user interview. But Futuur already had thousands of markets. The problem was never supply: people simply could not reach what was there.
The obvious fix was a personalized recommendation engine. It was out of reach for a small team at the time. So I set a different goal: remove the friction that cost users most, using changes the team could ship in weeks, and measure each one separately so we would know what actually worked.
Interviews showed two different moments. Some people arrive knowing what they want and need to find it fast. Others arrive open and need to be shown something worth their time. One feature cannot serve both, so I split the work into two systems and wrote one hypothesis per design improvement.
| Design improvement | Moment | Hypothesis |
|---|---|---|
| Filters | Find | If people can narrow thousands of markets by category and status, they will spend longer engaging with results and place more bets. |
| Search | Find | If search suggests markets before people type, fewer will give up on an empty box, and more will bet on the markets we choose to put forward. |
| Navigation | Browse | If categories are visible up to three levels deep, people will explore beyond the home feed. |
| Market cards | Browse | If cards are smaller and faster to load, people will see and open more markets per scroll. |
| Home banner | Browse | If the home page spotlights a timely event, more people will reach its markets and bet on them. |
Four deliveries, shipped one design improvement at a time.
Let people narrow the flood themselves.
Shipped the most-requested filters from interviews — category and status filters on the feed and on search — so users could cut thousands of markets down to what mattered to them. Alongside them, search gained a set of default suggestions, so people would not give up on an empty box.
- Engaged time, search
- +48%
- Recurring users, search
- +29%
- Engaged time, feed
- +23%
- Recurring users, feed
- −20%
Clear win on search; mixed on the feed — engaged time rose, but recurring users dropped. The default suggestions shipped with no data of their own: adding them was a design decision, since an empty search box gives people nowhere to start.
A new navigation architecture.
Redesigned the information architecture with up to three layers of categories visible at once — the single biggest lever we pulled.
- Pages per user
- +84%
- Engaged time per session
- +49%
- Bounce rate
- −12 pts
- Category pageviews
- +33%
Smaller cards, more content per scroll.
Simplified card design kept the essentials, showed more markets per screen, and loaded faster.
- Engagement rate, mobile
- +1 pt
- Engagement rate, desktop
- +4 pts
- Pages per session, mobile
- +4%
- Pages per session, desktop
- +3%
The quantitative data is inconclusive — too much changed alongside the new cards to credit them alone. But qualitative validations pointed the same way: users found the new cards easier to read. And the page got faster.
Spotlight the moment people already care about.
With Brazil's 2026 election approaching, we added a banner to the home feed pointing straight at the election markets — live odds, a countdown to the runoff, and one clear call to action.
- Weekly bets on the featured market
- +180%
- Weekly bets across election markets
- +102%
- Unique election bettors per week
- +81%
- New bettors, first purchase
- +80%
The first round was two weeks away, so part of the lift is organic interest. But the jump is a step in the banner's launch week, not a gradual ramp — and election bets grew +102% while the whole site grew +52%, lifting elections' share of all bets by a third.
Shipping one design improvement at a time is what made the results readable. Navigation and filters went out alone, so their numbers mean something. Market cards shipped alongside other changes, and all we can say is the page got faster. Next time, I'd protect a clean window for every change worth measuring, even when bundling is faster.
Deferring the recommendation engine paid off: four low-cost deliveries moved the metrics that mattered, and the team kept its capacity for harder problems. Personalization is still the next step, now with a clearer baseline to measure it against.