CodeshareHow we decide what to build next
Six products is already more than a small team should carry. What gets us to say yes, and the much longer list of things that get a no.
We build for communities: Discord servers, anime watchers, Minecraft players, people who want an AI that shows its working. Six products is already more than a team our size should comfortably carry, so the interesting question is not what we build. It is what we turn down.
What gets a yes
Something we would use ourselves, in a space we already understand, where the existing options are either expensive or bad. All six products started that way. None of them started from market research.
That is not a romantic preference. Building for a community you are not part of means guessing at what matters, and guessing produces features that demo well and go unused.
What gets a no
- Anything whose running cost grows faster than its usefulness. Free products have no headroom for that.
- Anything that needs constant manual attention. We are a small team; a product that needs babysitting takes the time the others need.
- Anything we would have to lie about. If the honest description is unappealing, the product is the problem.
- A seventh thing, when one of the six is being neglected.
The hardest call is stopping
Starting is easy and stopping is not. A product you have spent months on accumulates a case for itself: it might pick up, it is nearly there, it would be a waste to throw away the work.
The only reliable answer we have found is to decide on what is measurably true rather than on the story. When we closed the marketplace, the case for keeping it rested on numbers that did not survive being looked at properly. Once we had looked, the decision made itself.
The work is already spent. The only question is what the next month goes to.