People love an overnight success. Magic Seaweed was one — it just took a decade to happen overnight. I joined as founding editor in 2008, became a founding partner, and we built it into the world's largest surf forecasting platform before it sold for £13m. From the outside that's a clean headline. From the inside it was ten years of compounding, most of it invisible.
The premise was narrow on purpose: tell surfers, accurately, when and where the waves would be good. That's it. We weren't trying to be a lifestyle brand or a marketplace or a social network. We were trying to be the single most trusted answer to one question that a specific group of people asked obsessively, several times a day. Narrow is not a limitation. Narrow is how you become the default.
What made it work was patience, and I mean that as a strategy, not a personality trait. We built audience first and product alongside it, and we let both compound. A forecasting model gets better with data and time. Trust gets deeper with every accurate call. A community gets stickier the longer it's been someone's morning habit. None of those things can be rushed with a marketing budget — they accrue, slowly, and then they're a moat nobody can buy their way across.
That's the bit modern businesses find hardest to hear, because the whole culture is pointed the other way: launch fast, grow fast, raise fast, flip fast. Some things genuinely reward speed. But the most valuable thing we built — being the name surfers reached for without thinking — was the product of showing up and being right for nearly a decade. You cannot growth-hack your way into being a habit.
The exit, when it came, was its own education. Selling a company is not the romantic finale it looks like; it's diligence, and lawyers, and explaining a decade of decisions to people seeing them cold, and the strange grief of handing over a thing that has been part of your identity for a third of your life. 'Overnight' is doing a lot of work in that sentence. The headline number arrives in a moment; everything that made the number possible took years.
So when I talk to founders about 'patient product' now, I'm not being precious. I'm describing the most reliable way I know to build something genuinely defensible: pick a question you can own, answer it better than anyone for longer than anyone, and let the compounding do the work that no amount of spend can fake. It's slower than the pitch decks promise. It also actually works.