Apple, Google, Microsoft, Mozilla, and Igalia agreed on 20 focus areas for Interop 2026, continuing a five-year effort to improve cross-browser consistency. Fifteen areas are new and five continue from 2025. Progress is measured through automated Web Platform Tests and a public dashboard.
The list includes anchor positioning, contrast-color(), container style queries, cross-document view transitions, and scroll-driven animations. These features can replace categories of JavaScript and browser-specific work when implementations converge. Native animation timelines can also move work away from the main thread and avoid layout-driven scripting.
The operational gain is smaller bundles, fewer polyfills, and less manual cross-browser QA. It is not permission to delete every fallback on announcement day. The Interop score tracks tested behavior across vendors; a product still serves a particular browser mix, accessibility context, and set of older devices.
The dashboard measures vendors, not your traffic
The Interop 2026 score answers whether engines pass the Web Platform Tests for that area. Chrome, Firefox, Safari, and Edge sit on the same spreadsheet. The public number is useful because it is no longer a press release: you can see, area by area, what still fails.
It does not describe the audience your application serves. Corporate WebViews, older iPhone Safari, shared tablets, and the slow tail of sessions stay outside that average. A product with 8 percent of traffic on a lagging engine still pays the fallback cost. If that group concentrates a payment flow or a regulated channel, the cost is higher than the dashboard suggests.
The 2026 list is also not a single CSS block. It includes positioning, transitions, IndexedDB, the Navigation API, WebTransport, WebRTC, and JSPI for Wasm. Treating the announcement as "CSS is now standardized" mixes surfaces that move at different speeds. Anchor positioning, view transitions, scroll snap, WebRTC, and CSS zoom returned from 2025 because last year's slice still had failing tests. A carryover is remaining work.
Before removing a library, cross two sources. The first is the dashboard for the area you intend to use. The second is engine telemetry from the sessions that matter, including the tail of the distribution. Without that join, the team swaps a stable polyfill for a bug that only appears on a customer's device.
A polyfill leaves when the real audience has already moved
Floating UI, Popper, and scroll-linked animation packages exist because overlay placement and scroll timelines were not reliable across engines. When native support covers the case, the package stops paying for itself: updates, bundle weight, and layout conflicts.
Removal is a product decision, not a reflex to a vendor post. Define the minimum engine cut, measure the share of sessions below it, and only then take JavaScript off the critical path. Progressive enhancement remains the cheaper mechanism: native CSS takes the happy path, the script stays for anyone who still needs it. Runtime detection is worth more than a frozen user-agent table in the backlog.
contrast-color() shows the limit of an isolated feature. It returns black or white with better contrast against the given color. Safari and Firefox already shipped it; Interop presses the rest. The function does not compute contrast for text on an image, does not replace an audit, and does not pick a brand palette. Treating it as the end of accessibility work reopens the topic at the next review.
Scroll-driven animations on the compositor really do take work off the main thread when the implementation is correct. The visual effect, the preference for reduced motion, and behavior in nested overflow remain product tests. Dropping tens of kilobytes of JS is justified only if the same flow stays usable on the mix you support.
QA changes surface; it does not vanish
A heavy visual suite across four browsers usually exists because the same CSS produced four layouts. With less spacing divergence, full-engine screenshots lose value for some components. Interaction that a still image cannot see gains value: popover="hint" that does not close a menu, closedby on a dialog, a cross-document view transition, scroll snap in a keyboard-driven carousel.
Reallocate the time. Fewer padding regressions on a button. More flows with keyboard, screen reader, and reduced motion. Interop keeps accessibility and mobile-testing investigations outside the score of the 20 areas. Accessibility trees and dynamic viewports still lack the same institutional pressure as scored features. That belongs in the QA plan, not only in a CSS changelog.
I would stop opening the same static form in four windows. I would keep Safari on flows that use new overlay, navigation, and media APIs. A bug there costs a support ticket, not a pixel shifted in Figma.
Migrate by surface, with detection and rollback
A practical order lowers the risk of "we deleted the polyfill on Monday." Inventory packages that exist only to cover the 2026 areas. Rank each by bytes in the bundle and by damage if native support fails. Migrate first what has simple detection and a cheap fallback: style queries for theme, contrast-color() on a brand-colored button. Leave overlay, focus, and navigation last, where a miss breaks the main path.
Each migration needs three pieces of evidence in the pipeline: the engine cut from telemetry, a test of the flow with the native feature, and a way back (a flag or the package still installed) if the slow percentile fails. Without the third, the team finds the hole in production and rushes the library back in.
Advanced attr(), getAllRecords() in IndexedDB, and WebTransport may be irrelevant to your product. Do not spend cycle on them only because they sit in the 20. Interop aligns vendors. The local decision remains what the audience can already use, what still needs JS, and what QA should stop photographing so it can start exercising.
If your company is weighing something similar, reach me on LinkedIn or WhatsApp.
