Journal

Why we stopped using Tailwind for client work

We used Tailwind on every project for two years. Then we stopped. Here's what we learned about the difference between moving fast and building something that lasts.

For about two years, Tailwind was our default. Every project started with a Tailwind config. It was fast to set up, fast to prototype, and our whole team knew it. We shipped a lot of good work with it.

But over time, we started noticing a pattern. Client handoffs were hard. Anyone who wasn't already fluent in Tailwind had to learn a new mental model just to change a font size. The markup became dense — a paragraph with twenty utility classes is not readable markup.

What we use instead

We moved back to vanilla CSS — specifically, a small custom design token system with a handful of utility classes for layout. Nothing fancy. A spacing scale, a type scale, a color palette, and a grid. That's it.

The result: our CSS bundles are smaller, our markup is cleaner, and clients can actually edit things without our help. A developer who's never seen the codebase can open a stylesheet and immediately understand what's happening.

This isn't a hot take

Tailwind is a good tool. It's the right tool for a lot of projects — especially product teams who own their codebase long-term and have developers who know it well. But for client work, where we hand off to people who may not be developers at all, it creates more problems than it solves.

The lesson isn't 'Tailwind is bad.' It's that the right tool depends on who maintains the code after you leave.