Craft · 5 min read
Architecture commits. Digital iterates forever, and that reversibility quietly makes work worse, because nothing has to be right as long as it ships.
You can’t A/B test a building.
You can’t pour two foundations and keep the one with better engagement, or put the staircase on the left for half the residents and on the right for the rest and check the dashboard in two weeks. Once the concrete is in, the decision is made, and it will outlive you.
I spent eight years as an architect before moving into digital work. The biggest culture shock was discovering that in digital, nothing is ever finished.
The undo button
Every digital decision is reversible. Ship it, measure it, change it. It sounds like rigour, and mostly it’s a way of never having to decide.
Testing is brilliant for questions that have answers, like which subject line gets opened or which button gets clicked. It’s terrible for questions that need a point of view. A/B testing can’t give you an idea. It can only find the slightly less bad version of the one you already had.
Reversibility also changes behaviour long before any test runs. When nothing has to be right as long as it ships, the standard drops without anybody deciding to lower it. “We’ll fix it in the next iteration” is the most expensive sentence in digital, mostly because the next iteration rarely comes.
What commitment does to the work
Architecture is slow and painful partly because it commits, and commitment forces a particular kind of thinking. You have to know why every decision was made, because someone will ask, and the answer has to hold for twenty years. You test the idea on paper, in models and in arguments before anything gets built. And you design for the worst day, the storm or the crowd, instead of the average one.
The Gardener is the closest thing in my portfolio to a building. A film is also an object you finish and let go, and you can’t patch it after release. Its twist rests on a single cup: the old woman drinks from her worn dark mug, the crew get plain white ones, and the two tea scenes sound exactly the same. That only works if every other decision is locked, because one stray sound cue would give the whole thing away.
Bringing commitment back
You can’t pour concrete on a website, but you can borrow the mindset:
- Decide as if you can’t undo it. Undo it later if you have to, but make the first decision properly.
- Write down why. If the reason doesn’t fit on one line, you haven’t decided yet.
- Test ideas before launch. Stress-test the concept on paper with sceptics and save the A/B test for the button colour.
- Call something finished, out loud. It’s surprisingly hard, and surprisingly good for a team.
The trade-off
Reversibility is a gift: it lets digital teams move faster than any architect could. Without commitment, though, speed just produces more drafts.
The best digital work I’ve seen feels built. Someone decided and stood behind it.
The staircase is on the left. Live with it.