CASE STUDY — ケーススタディ
ト ー ナ メ ン ト シ ー デ ィ ン グ
I designed and helped build an automatic seeding tool for Super Smash Bros. Ultimate tournaments, used by organizers from college scenes to international majors.
ROLE
Product Designer & Frontend Dev
TIMELINE
6+ Months
STATUS
Shipped & Live
OUTCOME
3,657 events seeded in year one
概要 OVERVIEW
Smashbase.gg pulls tournament data from Start.gg and seeds Super Smash Bros. Ultimate brackets automatically. It handles player rankings, carpools and travel, separating players who keep running into each other, and overall bracket balance, in a fraction of the time it takes by hand.
問題 THE PROBLEM
Putting the best player at seed 1 is the easy part. A fair bracket also has to balance everything below, and it has to hold up across the whole bracket. Organizers used to spend their limited time on the top cut and leave the rest.
Online results were their own problem. The community mostly doesn’t count online SSBU results because of input lag, but players who dominate online aren’t beginners, and they’d demolish newer offline players. So an online set counts for 1% of an offline set. It’s in the algorithm, but it barely moves it.
SKILL
Seed by actual skill
Not by gut feeling or name recognition.
TRAVEL
Keep carpools apart
Players who traveled a long way together shouldn’t have to play each other in the early rounds.
HIDDEN TALENT
Strong players the local scene doesn’t know yet, like dominant online players, shouldn’t be badly underseeded.
VARIETY
At locals, players who face each other every week should be split up so the storylines and competition stay fresh.
NEW TO THE TOOLS
SELF-TAUGHT
FINAL SAY
We started from our own experience. We’d seen how a bad bracket draw at a major can cost a player the placement that sponsors and ranking committees judge them on. I wrote a discussion guide so the whole team could run consistent interviews, and we talked to TOs at every scale: international majors like Collision, Wavedash, and Tera in France, regional events, small college scenes across the US, and Team Mana, who run Smash scenes in the UAE, Saudi Arabia, and Qatar. TOs do this for the love of the community, so the tool had to feel like a trusted assistant they stayed in control of.
デザイン KEY DESIGN DECISIONS
A lot of TOs weren’t technical and were short on time, so a complicated settings screen would have failed. Seeding became a five-step wizard: pick the event, pick the tournament, adjust the starting seeds with drag and drop or numbers, set carpools and separation, then review and submit. No step needs technical knowledge. A TO can tap through every screen without touching an input and still get a high-integrity result, on a phone, under pressure.
The other big barrier was distrust of automation. TOs didn’t want a black box making calls they couldn’t explain to their community, so the review step shows a seeding diff: players who moved up are green, players who moved down are red, and arrows show how many spots each one shifted. A summary counts the swaps made for location, set history, and carpools. That fixed a real misunderstanding. Some TOs thought separation was broken because little changed, when the algorithm was choosing not to separate players where doing so would distort the bracket.
Drag-and-drop seeding came up again and again as a favorite feature. It was hard to build, and the easy move would have been to cut it. I pushed for it and did the research to help the dev team implement it.
アルゴリズム THE ALGORITHM
To earn trust, we ran the algorithm against human-seeded majors and shared the comparison publicly. It beat manual seeding on every integrity metric we measured.
Defaults adapt to the event, because integrity means something different at a weekly local than at a national major. Locals default to variety and splitting up frequent matchups. Majors default to strict skill-based seeding. TOs can override anything, but most don’t need to.
NO PLACEMENTS
Set results instead of placements
Placements get distorted by upsets, badly seeded opponents, and bracket luck. Set results are far more accurate.
SETS OVER GAMES
The set is what counts
For top-level seeding, games that didn’t match the set result are heavily discounted. Lower in the bracket the rule relaxes, which makes most of the bracket more accurate.
TIME DECAY
Old results lose weight over time, so legacy players can’t coast on past wins.
PRESTIGE
A win at a major counts for more than the same win at a casual local, where a top player might not be playing seriously.
ビジュアル VISUAL IDENTITY
We stayed away from the sterile, component-library look a lot of SaaS tools have, like default shadcn/ui. Standard component kits can feel alienating to non-technical users before they’ve even started, and a corporate look would have clashed with a grassroots community, and with who we are as a team. We wanted warmth and personality without giving up usability.
EVENTS SEEDED
3,657
Tournaments seeded in the year after the 2023 launch, from weekly locals to majors
MAJORS
10 / 36
Majors seeded with Smashbase in its first year
BIG ONLINE EVENTS
12 / 12
Every big online event that year was seeded with Smashbase
振り返り REFLECTION
Smashbase worked because we treated the community’s values as a requirement alongside the functional ones. The grassroots side of Smash is the reason TOs volunteer thousands of hours, and a tool that ignored it would have been rejected however accurate it was. The wizard, the seeding diff, the public stats: each one was there to make TOs confident enough to hand over a job they’d always done by hand.
TOs around the world told us Smashbase took a lot of the weight off running events, several said they’d pay real money for it, and one line kept coming up: “It saved our lives.” New organizers got the confidence to run their first events. There’s still work to do. Legacy players can be rewarded a little more than they should be, and we’re still tuning the time decay.