Consensium

Consensium is a debate platform for experts on topics like AI, economics, and ethics. You stake points on the answer you believe in. Your vote carries more weight based on your expertise and how much you put on the line.

I led UX and UI design on this project end-to-end — research, interaction design, prototyping, usability testing, UI, and the design system. The project ran through two distinct phases. In the second phase a product manager and a second product designer joined the team, and we worked side by side on the redesign.

I worked closely with the client, product owner, developers, and marketing throughout.

The original vision was a gamified platform built around staking points on opinions — covering any complex, debatable topic: AI, economics, ethics, and more. The goal was something rewarding and engaging, closer to a prediction market than a forum.

The space wasn't empty — it was broken. Prediction markets like Polymarket and Kalshi were intimidating to anyone outside finance. Debate tools were bland and confusing. Newer platforms were noisy and built for insiders. The gap was real: a place where backing your opinion felt natural, not overwhelming.

Three risks were on the table before any design work started. Empty City Syndrome: early users wouldn't find enough content to engage with. Sub-niche trap: too narrow an audience can't produce real consensus. Complexity: the staking mechanic had to be learnable fast — "complexity will kill" stayed in the conversation through both phases.

The client had an existing app flow to work from — the UX brief was to extend it, not reinvent it. The main challenge was the poll feed: how it felt to scan, how it communicated the staking mechanic, and whether it carried the right energy.

We explored a few directions and landed on something playful and colourful to reflect the gamified nature of the product, grounded in a clean white background to keep it breathable and approachable.

I led the visual direction and UI from there: defined the style, designed the core flows, and worked closely with engineering as we shipped.

When a new product manager joined, he ran a session with the client and the team to step back and look at the product honestly. The conclusion was uncomfortable but clear: the platform felt too playful for the audience we actually wanted — domain experts who treat their reputation seriously. The reference shifted from prediction-market to "Reddit or Stack Overflow, but for thoughtful debate."

Before changing direction, we tested the existing prototype with 5 potential users. The signal was consistent: people could complete the flows, but the tone felt off for the kind of conversations they wanted to have. Staking specifically felt like a casino mechanic to them, not a credibility signal.

The design sprint gave us a clear direction: build a social platform where people can discuss complex topics — not a debate tool, not a prediction market. A place where people engage with ideas and with each other.

We started with the social essentials: profiles, points, and levels. The feed became scrollable and minimal — less noise, more content. One thing to read, one thing to do.

The staking mechanic was the core differentiator — but the experience around it felt foreign. "Stake" didn't land. Users understood voting, not staking.

We reworked the whole flow: instead of staking, users push an answer up or down — and can withdraw their support at any time. Points earned became Reps — reputation that reflects how credible your contributions are over time. The mechanic stayed the same, but the language matched how people already think about content.

After testing with 10 users on UserBrain, they understood it without explanation.

In a brainstorm session on levels, a key decision emerged: staking shouldn't be available to first-time users. New users need to experience the discussion layer first — ask, answer, engage — before the credibility mechanic unlocks. This protected the quality of staking and gave new users a clear path in.

Onboarding was one of the most iterated parts of the project. We tested several approaches with different goals — prompting users to ask a question, prompting them to answer one, a general walkthrough with platform rules, and one with example questions.

To support different acquisition channels, I designed five landing pages: a general marketing page, a business-facing page, a hackathon page, and two variants for paid campaigns.