A relaunch is a six-month bet placed on one day

A typical relaunch freezes the live site for months, then ships everything at once. New design, new copy, new navigation, new structure, all on the same morning. Six months of decisions, placed as a single bet.

I have run agencies that sold these projects, and two things went wrong often enough to count as a pattern. First, you cannot tell which change did what. Traffic moves, enquiries move, and twenty changes landed together, so the result belongs to all of them and none of them. Second, reversing one part is expensive. The old site is gone, the team has moved on, and untangling a single decision from the rest is a project in itself.

The alternative I now work with is one change per week, live by Friday. A shorter form. A different headline. A new layout for one page. Each one small enough to judge on its own and cheap to take back if it is wrong.

Small steps only work when they point somewhere. So the first week is not a change. It is one file that describes what the site should be in six months and why. Every weekly change then has to move the site toward that file, and the file can be revised as the work teaches you something.

I wrote the plan for atonda.com on 27 October, one item per week, About page first, because that page no longer described what I actually do. It took five versions of the copy to get right. The homepage is still the old one.

Week one shipped four things. The new About page. A redirect that had been broken for months. Footer links that pointed to the wrong places. Click tracking on the contact links.

A week of numbers on a personal site tells me very little, and click tracking shows that someone used a link, not that they wrote to me. What I can already say is that the links work, the About page describes my work more accurately, and nothing needs rolling back. That is more than I would know a week into a relaunch.

Some changes cannot be separated. An end-of-life platform. A framework that cannot be upgraded in place. A structure that genuinely needs rethinking. Those need a bigger move. Even then, the version that works is the smallest one. Migrate what exists, change nothing else, and start the weekly improvements on day one. The migration that quietly grows into a redesign of copy and structure is the one that becomes the six-month bet again.

Not when the relaunch is. What ships next Friday, and how will you know it worked. A team that can answer that will hand you a better site in six months than a team promising a reveal.