De 8 bases de données à 800 en 3 jours après un lancement accidentel sur Hacker News

The post was not a deliberate launch: an early GitHub follower submitted it while the team was still alpha testing, months before their planned official launch.

“the HackerNews post was created by an early GitHub follower, while we were alpha testing”

Paul's own read, written days after the launch, is that shipping in public early let them catch a moment they could not have planned for. In a 2025 retrospective (third-party hosted, so implied confidence) he adds that the timing coincided with a tagline change from "real-time Postgres" to "open-source Firebase alternative", which he says is the positioning that then stuck.

“It just goes to show: ship early and often because nobody knows when you might get picked up.”

After finishing YC, the team recreated a demo-day-like deadline for themselves every few months, then escalated it into shipping one new thing every day for a week.

Paul attributes a measurable top-of-funnel lift to the very first Launch Week, which is why the format became a recurring, several-times-a-year event.

Their operating model was to build on top of existing open source projects rather than reinvent them, upstreaming changes and directly employing maintainers of the tools Supabase depends on.

“We want to leverage existing open source products wherever we can, improving them by up-streaming changes and employing maintainers”

By the Series A announcement (Oct 2021) Paul is explicit that none of this growth came from paid acquisition.

“Our growth is all organic.”