A better feed is still a feed

Most of my portfolio is trucks, dispatch boards and enterprise tables. So naturally the project that taught me the most about product strategy was a place to play Magic: The Gathering on a livestream. With real cardboard.

Across 2025 and into this year I was the only designer on a digital arena for the trading card game community: a platform where players stream and play with their actual physical cards, across more than twenty games, from Magic to Pokemon to Yu-Gi-Oh. A consumer platform of fourteen-plus modules, and an admin panel of seventeen-plus screens behind it.

The most important decision on it was a thing I refused to build.

Five to seven tabs for one game of cards

The trading card game community is large, active, and organised entirely on tools built for somebody else.

A player who wants to stream a game runs it through a streaming platform where card games sit buried beneath millions of mainstream streams. Coordination happens in chat servers, by hand. Discussion lives in threads and groups and ephemeral chat that nobody can search a week later. Creators build audiences on platforms whose algorithms favour mainstream gaming. Finding a local store is a map search that knows nothing about the rest of it.

Running a single game session took three or more tools. There was no such thing as a lobby for physical card games, no way to browse open sessions by game, format, player count or privacy. And spectating, the thing that makes tournament play exciting, was absent from everyday games entirely, because there was no viewing experience separate from the player's own.

The brief was to consolidate all of it into one arena. That brief contains a trap.

The obvious architecture

Every platform in the adjacent space is organised around a feed. Streaming sites, video sites, every social product. A content-first architecture with the feed at the centre was the default answer, and honestly, the client's own creator ambitions would have justified it.

I argued against it. A feed-centred platform competing with the largest streaming and video platforms on their own terms is a fight over recommendation algorithms and content volume. A startup loses that fight, every time, and it loses it slowly enough to spend all its money first.

So the question changed from "where are these platforms weak for card game players" to "what can none of them host at all". The answer was the game.

Everything hangs off the table

The live game session became the structural object every module attaches to. Streams, spectators, forums, creator content and team profiles all derive from play or point back to it, instead of sitting beside it as separate products that happen to share a login.

That also fixed discovery in the right direction. People arrive for a game and find the community, rather than arriving for content and being asked to also play.

The parts with no precedent took the longest and ended up holding the product's identity. The game lobby is a browse, filter, create and join model for physical card games, with sessions carrying the host, viewer count, open player slots, and join, spectate or request actions. Gameplay supports up to four versus four with commanders, decklists and customisable counters. And spectating is its own experience with its own stream view and chat, designed to hold thousands of concurrent viewers, not a read-only copy of the player's screen, because watching a game and playing one are different jobs.

Three smaller calls, same logic

Once the centre was fixed, a lot of smaller decisions got easier, because each one could be judged by whether it made the table busier.

Registration became personalisation rather than a gate. Four steps: account, avatar, which games you play, and your player card. The metric that matters on a new platform is not accounts created, it is accounts that come back, and the classic failure is a new user landing on an empty dashboard. Asking which games someone plays is not friction. It is what makes the first home screen worth looking at.

The privacy choice shows its cost at the moment it is made. When a host sets a game to private, a confirmation tells them that public games typically receive five to ten times more player engagement. A cautious host is making a reasonable individual decision with a cost they cannot see, and that cost lands on the platform as an empty lobby.

Creator curation pairs editorial and algorithmic ranking in one interface. Manual curation alone quietly becomes a popularity contest among creators the admin already knows. Full automation surfaces whoever is already largest. So the spotlight stays hand-picked, and the new and rising slots run on criteria, because those are the places where fairness matters more than taste.

The honest part

I do not play Magic.

A platform for people who do is full of decisions that look arbitrary from outside and obvious from inside. Which formats belong in a filter. Whether a commander is a field or a first-class object. What a player card should actually say. The founders are players, so their review carried real weight, which is a better position than most engagements offer. It is still not the same as watching a stranger use it, and I would not mistake one for the other.

And not every module is a triumph. The ones I could pattern-match, forums, messaging, profiles, are competent and unremarkable. That is the right outcome for them. Nobody joins a card game platform for the direct messages.

What it launched with

More than twenty supported games from day one. The client cites an active base of over 1,400 players, 500 streams and 2,000 hours of content, on an architecture intended to scale toward 12,500 users and creators, with both applications sitting on a custom design system I built from scratch, light and dark themes included.

The lesson I kept

My first instinct on a platform competing with giants was to find where each one is weak for this audience. That is a real list, and a losing one, because a better version of what they already do is still a version of what they already do.

Look for the thing incumbents structurally cannot do, not the thing they do badly. A better feed is still a feed, and they have more of it.