How we think
Four principles, and what each one costs
Every one of these makes something harder. We keep them because the thing they make harder is cheaper than the thing they prevent.
- 01
Design for the second language, not the first
Costs: Slower for the first month
A product that speaks one language and a product that speaks five are not the same product with a translation file between them. They differ in the schema, in how search is indexed, in how a price or a date is rendered, and in how much room a button needs. Deciding that on day one costs a slower first month. Deciding it in year two costs a migration with a downtime window.
- 02
Ship the smallest honest version
Costs: Says no to features you asked for
Every product on our work page went live as an MVP that did less than the brief described. That is deliberate. A feature nobody has used yet is a guess, and a guess in production is more expensive than a guess in a document. We would rather cut two things and learn which one you actually needed.
- 03
The people who build it carry the pager
Costs: Nobody here gets to stop thinking about operations
We run our products on infrastructure we administer. That arrangement is not efficient, and it is the single best quality control we have: a shortcut taken on a Thursday is a shortcut that wakes its author on a Saturday. Nothing focuses a design review like knowing who answers when it breaks.
- 04
Build it so you could leave
Costs: Documentation nobody is forced to read
A product that only its authors can deploy is not finished, it is rented. We write for the engineer who was not in the room, and we would rather hand a project over cleanly than keep it by being the only people who understand it.
The rest of this is easier to show than to explain.
Four products are live in Korea right now. Read what each one had to solve, then tell us what you are trying to build.