Why we only build in one CMS

We turn down work every month because it's WordPress, or Webflow, or Shopify. It's the least commercially sensible decision we've made and the one we'd defend hardest, so here are both halves: what it buys, and what it costs.

Estimates stop being guesses

When you build the same kind of thing repeatedly, you stop estimating and start remembering. We know what a Matrix-heavy marketing site costs because we built one last month, and one before that, and one before that. The variance between them is small and we know which factors drive it.

That's the whole reason we can quote fixed prices and publish a rate card. It's arithmetic rather than nerve. The uncertainty is smaller, so the risk premium can be smaller too.

A generalist quoting across four platforms is carrying four separate sets of unknowns, and has to price for the worst of them. That risk premium is real and it comes out of someone's margin, usually the studio's. It's the hidden cost of a partner who can do everything.

The patterns compound

The second build in a system is faster than the first. The tenth is much faster, and not because anyone is typing quicker.

It's because the problems are already solved: how to model content so it stays editable, how to structure a block library so it doesn't sprawl, how to handle the parts of an upgrade that always go wrong, how to set up deployment so a launch isn't an event. Each of those took time to work out once, and now costs nothing.

Spread that across four platforms and you have four shallow versions of the same knowledge, each one re-learned a bit whenever you come back to it, none of them deep enough to be worth much.

Support stops being archaeology

Because we host most of what we build, and it's all the same platform, a support request in year three is a question about a system we know well. There's no re-learning phase, no working out which generation of somebody's conventions this site belongs to, no discovering that the previous developer solved a problem in a way we've never seen.

It also means any of us can pick up any site. A studio doesn't have to wait for the one person who remembers their project, which is exactly what makes small agencies painful to work with when something urgent happens.

Doing five things adequately is worth less to a studio than doing one thing without surprises.

We get better at the parts nobody sees

Specialising has a second-order effect that took us a while to notice. Because we aren't spending attention on breadth, we spend it on depth: on the control panel being well organised, on the block library being coherent, on the upgrade path being clean.

Those are the things a client experiences daily and a studio never sees, and they're the first things to get skipped when a team is stretched across platforms and just needs the thing to ship.

What it costs us

Work, regularly, and it's worth being clear about the size of it. A studio arrives with three projects, one of them Craft. We take one and decline two, and sometimes lose all three, because they'd reasonably rather have a single partner for everything than manage two relationships.

That's a real loss, it happens often enough to notice, and we've made peace with it rather than solved it.

There's a second cost that's harder to talk about. Our fortunes are tied to one platform's. If Craft stopped being well maintained, or changed direction in a way that didn't suit the work we do, we'd have a serious problem and not much of a hedge. We watch that closely, we think the risk is low, and it wouldn't be true to say it's zero.

What it buys you

Estimates that hold, a rate card you can quote from on the day, a short handover (because the conventions are conventional), support from the people who wrote it, and no learning curve billed to your project as a slightly larger number than it needed to be.

That's the trade, and we think it's a good one. A version of us that claimed to be good at everything would be a lot less useful to you.