Most websites get commissioned as a brand exercise and end up as infrastructure. Nobody plans that transition, which is why it usually goes badly.
The commissioning conversation is about positioning, tone and design. Reasonable — that is what a new site is for. But six months later the same site is where enquiries arrive, where a proposal gets sent from, where a client checks a delivery date, and where three people go to find the current version of a price. It has quietly become a system that the business runs on, still built as a brochure that someone updates.
The tell is a queue. Somebody wants a page changed and cannot change it, so they ask, and the request joins a list. The list is the bottleneck, not the site. Every day that page is wrong is a day the business is describing itself inaccurately, and the cost is invisible because nobody attributes a lost enquiry to a stale page.
We have watched this play out as a specific pattern: the offer moves faster than the site can. A studio changes what it sells, or how it prices, and the site keeps advertising the previous version for a quarter because editing it requires a developer. At that point the site is not just failing to help, it is actively misinforming.
So the question we ask early is not what the site should look like. It is what has to stay editable, and by whom. That single answer reshapes the build: which parts are content, which are structure, who owns each one, and what happens when the offer changes again.
The second question is what the site has to be able to do rather than say. Take an enquiry and route it somewhere a human will see it. Show a live number instead of one typed in last March. Let a client see the state of their own job without an email. Those are operating requirements, and they are cheaper to build in at the start than to bolt on once the structure has hardened.
This is also why we write content to the structure rather than pouring it in afterwards. Copy written for a layout that does not exist yet gets cut to fit; copy written into a structure that knows what it is for tends to survive.
None of this means the brand surface does not matter. It matters enormously — it is the first thing anyone judges. The argument is only that the two jobs are not sequential. A site that looks right and cannot be maintained becomes wrong within a year, and a site that is easy to maintain but says nothing was never worth visiting.
The version of this that works is unglamorous: fewer pages, clearer ownership, and a structure that assumes the business will change. The site stops being a project that finished and becomes a thing that runs.