Inspired by a recent piece from Ian Styles on why startups should build brands in beta rather than treating them as finished artefacts.
A design studio recently made a claim I haven't been able to shake.
Most startups, they argued, try to launch a finished brand. The logo is locked. The language is fixed. The rules are written. Everything is polished, approved and frozen into place.
The problem is that nothing around it is frozen.
The category is still forming. The language keeps shifting. The product evolves. New competitors appear. The market discovers what it actually cares about. By the time the brand has settled into itself, the world it was built for has already moved on.
Their solution was simple: don't build a finished brand. Build a brand in beta.
Not a loose collection of assets, but a system of reusable parts. Type, grids, shapes, components and principles that can be recombined as the company grows. Something coherent enough to create consistency, but flexible enough to survive contact with reality.
Ian Styles was writing about branding.
I think the idea reaches much further.
The thing you freeze is the thing that breaks.
Most people assume uncertainty lives in the outcome. In my experience it usually lives in the route.
The outcome is often surprisingly clear. Raise the fund. Detect the deepfake. Bring together the right seventy people. Build trust. Move capital. Change a policy. Launch the product.
The uncertainty begins the moment you start moving towards it.
Markets shift. Platforms change the rules. A key hire leaves. A donor pulls out. A technology that looked inevitable turns out to be a dead end.
The mistake isn't having a plan.
The mistake is becoming emotionally attached to one.
What survives over time is not rigidity but a particular kind of discipline: identifying what is genuinely essential and refusing to compromise on it, while allowing everything else to adapt as conditions change.
That's true of brands.
It's true of institutions.
It's true of strategy.
And increasingly, it's true of people.