Kanga

Kanga was a viewer-first platform for online creator communities that aimed to improve content discovery and aggregation, livestream engagement, and creator monetization. I joined as Lead Designer at the start of 2019, when the product was an unnamed development build. Two and a half years later, Riot Games acquired the company. This case study is about the strategy in between: shipping the wrong first product, reading the data honestly, and pivoting to the thesis that made the second one work.

Context & challenge

Livestreams are scattered across many platforms, including Twitch and YouTube, which lead to challenges in moment-specific discovery: it was hard to find something worth watching right now. That was the problem the founding team set out to solve with an index of streams enriched with context via computer-vision technology. My role as the sole designer was to take our powerful content engine and shape the product around it.

The wrong first product01

When I joined, the team had put together a development build that was functional but lacking in clear information architecture or visual polish. My first few months with the team were focused on foundational work, including working with a branding agency on the product name, logo, and brand identity. A few months in, we launched our first version of Kanga.gg as a Chrome extension and mobile app using React Native.

What the data said

The launch was successful, with thousands of initial users in the first few weeks, but the metrics had strong signal that our product was on the wrong track. Engagement was very high but retention was low. We watched that pattern hold for weeks while we ran user interviews and iterated on the product. Ultimately, we identified that our solution was too niche. We had built a sharp tool for a narrow market need, but we needed to build a mass-market product. Better content discovery was a feature, not a company.

The pivot to collaborating with creators02

Two findings from those interviews converged into the pivot.

The first was about viewers. The parasocial relationship between viewers and creators was very clear. Users cared a lot about the creators they followed, and a creator's endorsement of a product carried a lot of weight when it came to trying new platforms and experiences.

The second was structural. The richer experiences we wanted to build, like interactive experiences tied to a live broadcast, required the creator to configure and run software on their stream. We couldn't build for viewer experience anymore without building for creators first.

So the creator became our market wedge. We served creators with engagement and monetization tools, and their audience would follow them to Kanga where we could serve audiences with richer, easier experiences. The two sides of the flywheel converted into real growth: collaboration with creators drove hundreds of thousands of organic visits and a steady 2% account-creation rate.

Measuring product–market fit on both sides

We rebuilt our goals framework around the two-sided market: 28-day retention and on-platform engagement on the audience side, and creators onboarded plus an NPS-style approval score on the creator side.

Defining and tracking this was a collaborative effort I helped shape. I treated measurement as part of the design process rather than a post-launch check, a practice that runs through everything I've shipped since.

Designing the platform03

Onboarding had to move a viewer from zero to a personalized experience in one session, so it synced a user's Twitch follows on sign-up. The home architecture organized around immediacy: Live Now, what we recommend, and who you already follow.

The engagement layer gamified the live broadcast with real-time predictions and auto-resolving prop bets on in-game events that gave viewers a stake in the moment. Iteration was rapid and we shipped updates weekly.

The acquisition04

In summer 2021, Riot Games acquired Kanga. We brought everything we made and everything we learned to Riot's Esports department in the pursuit of building the first all-inclusive platform for all of Riot's content: Riot Esports Network.

The acquisition validated the thesis: increasing fan engagement is key to growing the content economy and establishing esports properties as generational media.

Reflection05

Engagement is not retention. Our first product had enthusiastic users and no staying power. Learning to distinguish a spike from a habit and call it out before the sunk costs grew was the most valuable thing the failed version 1.0 taught me.

The market tells you where to build, if you ask early enough. We reached the creator pivot through weeks of interviews after launch. The honest critique is that the same interviews before launch would have gotten us there a quarter sooner. That lesson matured into the data-led approach to design I still use today at Fizz.

A two-sided product needs two-sided metrics. Optimizing only the viewer side had us polishing the wrong surface. Splitting the goals framework across creators and audiences let us draw insights that uniquely fit each persona and changed what we shipped.