Building for scale: lessons from launching in three new markets.

We launched in three markets in the same quarter, which in hindsight was either brave or reckless depending on the week you asked us.

The infrastructure scaled the way infrastructure usually does — with a few uncomfortable nights and a lot of very boring, very necessary capacity planning. The harder problem was never the servers.

The harder problem was that each market needed a slightly different product before anyone would admit it out loud. Language wasn't a translation problem, it was a trust problem. People didn't want a platform that had been translated at them — they wanted one that felt like it had been built with them in the room.

The best products solve problems people can't articulate yet. Nobody in any of the three markets asked us for what we eventually built. They just kept telling us, in a dozen indirect ways, what wasn't working — and our job was to hear the problem underneath the feedback.