CSS, for anyone who hasn’t touched it since a school computer class, is the code that tells a webpage how to look: where things sit, what color they are, how they move. It doesn’t run the logic of an app; it dresses it. And because the web isn’t owned by one company but rendered by several competing engines — Chrome’s, Safari’s, Firefox’s — a CSS feature can be finished, documented, and beloved, yet still unusable in a real project if even one major browser hasn’t caught up.
The feature everyone wants and nobody fully trusts
This year’s clearest example is anchor positioning, a feature that lets developers pin things like tooltips, dropdowns, and popover menus relative to other elements without resorting to JavaScript workarounds. In the survey, it topped the list of favorite new features — tied closely with :has() — and it also topped the list of features developers most want to use but avoid because of browser support gaps. That’s not a contradiction so much as a portrait of where CSS stands right now: excitement and hesitation, aimed at the exact same feature.
The CSS-Tricks team, summarizing the results, put it plainly: anchor positioning "scratches an itch that’s annoyed many of us for years," doing for tooltips what used to require careful JavaScript juggling. But loving a tool and shipping it to real users are different acts, and the survey suggests many developers are stuck in the gap between the two.
A companion analysis of the same data frames this as the defining story of the whole survey: "enthusiasm for newly-announced features is often mitigated by the inability to use them in a real-world production context". Other features fall into the same bucket — view transitions (for smoother page-to-page animation), the conditional if() function, and scroll-driven animations all show up as things developers want but currently can’t rely on everywhere.
Old favorites, new complaints
Interestingly, some of the survey’s "favorite" features aren’t new at all. :has(), container size queries, and CSS nesting have all had support across major browsers since 2023, and Grid has been universal since 2017 — yet they still rank among developers’ most-loved tools. One reasonable reading, echoed in the survey analysis, is that adoption simply lags discovery: it can take a few years for a feature to go from "available everywhere" to "something teams actually reach for by default". Familiarity, in other words, doesn’t kill affection — CSS developers just take their time falling in love.
That lag cuts both ways, though. Grid and older color tools are mature enough that changing them now is difficult, and the survey’s free-text comments show real frustration with things that shipped years ago — Grid’s syntax is called "unintuitive," and relative and wide-gamut color workflows are described as verbose and hard to remember. Mature doesn’t always mean comfortable.
Here’s a condensed view of how the pieces fit together:
| Category | What it shows | Example from the survey |
|---|---|---|
| Favorite new feature | What developers are most excited about | Anchor positioning, tied with :has() |
| Browser-support pain | What developers want but can’t reliably use | Anchor positioning again, plus view transitions and if() |
| Top missing feature | What developers wish existed natively | Mixins — reusable style logic, borrowed from preprocessors like Sass |
The overlap between the first two rows is the heart of the story: the very feature people are most excited about is also the one most held back by inconsistent support.
Why a feature takes years to become "safe"
It helps to picture the adoption path a CSS feature typically travels — not because every feature follows it exactly, but because it explains why hype and production use rarely arrive together.
flowchart LR A[Feature announced] --> B[Developers experiment] B --> C[Uneven browser support] C --> D[Progressive enhancement as workaround] D --> E[Interop efforts narrow gaps] E --> F[Feature reaches Baseline]
That "progressive enhancement" step is doing quiet, essential work. It’s the practice of shipping a feature so that browsers which support it get the nicer experience, while browsers that don’t simply fall back to something plainer — nobody’s site breaks either way. Industry efforts like Interop exist specifically to speed up step three by getting competing browser makers to agree on shared priorities. None of this guarantees a fast timeline; it’s more of a mechanism than a schedule, and the survey itself offers no promise about when any particular gap closes.
CSS’s quiet resistance to AI
There’s a second, less obvious storyline buried in the numbers: while AI-assisted coding has spread through much of software development, CSS looks different. Among survey respondents, 3,031 reported using AI generation somewhere in the 0–50% range of their CSS work, compared with only 721 reporting usage above that range. That’s a self-reported pattern from this particular group of respondents, not proof that AI performs worse at CSS everywhere — but the comments accompanying the numbers lean the same direction, with many respondents saying AI tools simply aren’t yet producing CSS they trust. CSS, more than most corners of web development, still seems to reward — or require — a human hand.
A satisfaction score that hides some complexity
Despite all this friction, developers rate their overall happiness with CSS at 4 out of 5, unchanged from last year and part of a gradual upward drift since 2020. That’s a curious number to sit next to so much visible frustration, and it’s worth resisting the urge to read it as proof the ecosystem has become simple. It’s more likely evidence that CSS has become dramatically more capable — able to do things that once required JavaScript hacks or design compromises — even while the seams between browsers still catch people. Powerful and occasionally exhausting aren’t mutually exclusive, as the CSS-Tricks writeup wryly noted.
The survey also surfaces a less comfortable pattern: a persistent gender and ethnic imbalance among respondents, described in the source material as "as woeful as ever". The data can’t tell us whether that reflects who works in CSS broadly or simply who chooses to answer a survey like this one — but either reading points to the same practical conclusion, that broadening participation will take deliberate effort from conferences, employers, and publications, not just repeated surveys.
What this all adds up to
None of this makes anchor positioning, :has(), or any other favorite feature "ready" in some universal sense — popularity in a survey isn’t the same as production-safe everywhere, and this pool of respondents, however large, isn’t a stand-in for every person who writes CSS. What the 2026 results do offer is a clearer picture of how a web technology matures: not through a single announcement, but through years of overlapping excitement, friction, workaround, and slow interoperability gains. CSS in 2026 isn’t stuck — it’s just moving at the deliberate, uneven pace of a platform that has to convince several different browsers to agree before a good idea becomes something you can simply use.
