Lea Verou
We often expect that the flashiest features would attract the most developer interest. Things like AI-powered APIs, 3D/XR/AR, or device APIs. Yet year after year, what rises to the top is anything but fancy. Almost… mundane: Dropdowns. Comboboxes. Popovers. Dialogs. Form validation. Loading and saving files. Templating. Safely displaying user-generated content. Icons.
Making APIs Usable
“But — many of these already exist!” you might object. They do — but when built-in UI can’t be customized or styled, it is effectively unusable. Developers are forced to recreate it anyway.
Once that penny drops, it explains most of the survey results. It explains why Customizable Select topped the charts. Why low-level primitives like popovers and invokers are so well-received. Why SVG and drawing HTML on canvas ranked far above AR/VR. Even why extending built-in elements emerged as such a major Web Components pain point.
After all: how many times can you rebuild a button or dropdown before losing your mind? And how confident are you that you got it right?
As a result, building professional interactive UIs still requires cobbling together numerous third-party solutions, even for things that are routine in proprietary platforms.
Lagging Change
This finding is not new. In the first State of HTML, we already saw how critical styling and customization were, especially around forms. This year’s deeper analysis confirms it hasn’t changed — in fact, the signal is louder.
But things have changed for the better! Features that were being discussed in 2023 have now shipped in major browsers. Others that were impractical back then—like the Popover API—are now universally supported.
And yet, Popover still tops browser support complaints. Why? While the pace of implementation has accelerated, collective perception lags behind. It now takes longer for developers to trust support than for browsers to ship it!
The Impact of LLMs
The rise of AI-generated code is a contributing factor. LLMs are surprisingly well-versed in modern web platform features, and could have helped move the community forwards, but are overly conservative in what they recommend. In practice, this may even be slowing adoption! How to fix this at scale remains an open question. Meanwhile, explicit browser support instructions do help, and who knows, if enough devs do this, the problem may even fix itself.
Looking Ahead
Looking ahead, the future looks bright. Many of the pain points highlighted are actively being worked on, and in some cases already shipping. Even extensible built-ins, previously considered a dead end, is seeing potential progress through the early stage work around custom attributes.
All these primitives point in the same direction: a web platform that is steadily becoming more flexible, more expressive, and more aligned with how developers actually build UIs.
Progress on the Web is rarely dramatic, but it is cumulative. Each new primitive reduces the need for workarounds, libraries, and fragile hacks. And when those fundamentals fall into place, the impact will be felt everywhere.