The feature that sold most teams on Sass — nesting — is now built into CSS. That does not make Sass pointless. It makes the decision narrower and more interesting.
What changed in CSS
Native nesting, custom properties, calc(), color-mix(), and cascade layers cover a large share of what preprocessors were used for. A modern stylesheet can nest selectors and keep a color token in one place without any build step.
The crucial difference is when things happen. Sass variables are compiled away; the browser never sees them. CSS custom properties survive into the runtime, which means they can respond to media queries, :has(), or a value set from JavaScript.
Side by side
| Sass (SCSS) | Native CSS | |
|---|---|---|
| Variables | compile-time only | runtime, cascade-aware |
| Nesting | yes | yes, in current browsers |
| Mixins | yes | no direct equivalent |
| Loops and functions | yes | no |
| Build step | required | none |
| Browser output | plain CSS | is the CSS |
SCSS:
$brand: #0d9488;
.card {
color: $brand;
&:hover { background: #f0fdfa; }
}
Compiled CSS:
.card {
color: #0d9488;
}
.card:hover {
background: #f0fdfa;
}
Native CSS, same result, no build:
:root { --brand: #0d9488; }
.card {
color: var(--brand);
&:hover { background: #f0fdfa; }
}
What Sass still earns
Mixins with arguments, loops over maps, and functions that generate values have no native replacement. If you maintain a design system with dozens of spacing tokens, theme variants, or repeated media-query patterns, a preprocessor still removes real repetition.
Small sites are the opposite case. A 300-line stylesheet does not need a compiler in the deploy pipeline; native nesting and custom properties are enough.
Compiling without a build step
No compiler installed? The Sass to CSS converter runs dart-sass in your browser — paste SCSS or indented Sass and get standard CSS out, nothing uploaded. It handles both syntaxes, so it doubles as a quick sanity check for syntax errors.
For projects with a build pipeline, the CLI remains the standard tool:
npx sass input.scss output.css
Pick by maintenance cost, not fashion: if the build step saves more repetition than it creates, keep it. If it only exists because it did in 2018, delete it.