Maintenance
Longevity as a design principle
Brittle is a slow death sentence.
The average website lasts about two years and four months before someone tears it down and starts over. Most of those rebuilds aren't about taste. The site didn't get ugly. It got brittle, and brittle kills a site on a slow timer you can't see running.
Open the plugin list on most sites and you see it: a few badges green and current, the rest flagged not updated in two years. That quiet pile is nobody's job to review, and it's the thing that decides whether your site is still standing when you need it.
When we're building, the question isn't what's the fastest way to add this carousel. It's are we willing to own this carousel in three years, when the plugin author disappears. Most of the time the answer is no. So it goes in as fifteen lines of code we actually understand, instead of a four thousand line plugin rented from a stranger. Less borrowed code is less code that can rot out from under you while you sleep.
Three shifts that keep a site standing.
Every dependency is something we have to own in year three. If we won't own it, it doesn't go in. We understand every part we ship.
Boring tools have a property exciting ones don't: a giant pile of people already know how to fix them. When your site breaks in year four, you don't want it breaking in a language nine people understand.
Most sites die of bloat: campaign pages, microsites, sections for events two years gone. Longevity means having an opinion about what this site is allowed to become, and holding it.
Every dependency is a debt, not a feature.
A feature you rent looks free on launch day and comes due the year the person who wrote it moves on. We'd rather owe nothing we can't pay back ourselves.
The part that doesn't demo well.
Because we chose not to use the fancy plugin is a terrible slide in a pitch deck, and choosing for the room is exactly how you get a site that dies at two years and four months. We optimize for year five, not for the demo.
A maintenance contract manages the problem after it is built in. Designing for longevity prevents it, the same discipline behind Keystone and behind how we scope every build from the start.