We used to build client sites on page builders. We stopped in 2019, and the reason was not performance, or snobbery, or any of the things people usually assume.
It was the third time in a year that a client asked us to change something small — a heading style, sitewide — and the honest estimate was eleven hours.
In a custom theme, a heading style lives in one CSS rule. Changing it everywhere is one line and five minutes, including making the tea.
In a page builder, that same heading exists as a set of options on every module on every page, serialised into a database row. There is no single place to change. There is a global setting, which the client’s team has overridden on forty pages because the global setting did not do what they wanted on that one occasion.
This is what makes it hard to argue about at the start of a project. A page builder genuinely is faster for the first three months. The site gets built quicker and the client can move things around, which they like.
The cost turns up in year two, and it turns up as a series of small estimates that feel unreasonable. Nobody connects them back to the decision that caused them.
We build a custom theme and register a block for every layout in the design. The client’s team gets the same freedom to assemble pages — arguably more, because the blocks are built for their content rather than being general-purpose — and the styling lives in the theme where it can be changed once.
It takes us maybe fifteen percent longer to build. It has never once cost a client eleven hours to change a heading.
If the site is genuinely temporary. A campaign microsite with a three-month life has no year two, and everything above is irrelevant. We have built those on page builders and would again.