The Core Web Vitals thresholds haven't moved since 2024. LCP under 2.5 seconds, INP under 200 milliseconds, CLS under 0.1. Nobody can claim the bar moved on them, and yet more than a quarter of sites are still failing the metric that's been around long enough to stop being an excuse.
The Numbers Didn't Change. The Consequences Did
Following the March 2026 core update, pages in position 1 on Google show a 10% higher Core Web Vitals pass rate than pages sitting in position 9. Core Web Vitals were never a hard ranking override, they're a page-experience signal, but that gap is now a visible pattern in who's winning. Passing them stopped being optional the moment ranking pages started passing them at nearly double the rate of the pages below them.
INP Is the One Nobody Fixed
28% of origins are still failing INP (Interaction to Next Paint), two full years after it replaced FID as the interactivity metric. Unlike LCP, which is often one bloated hero image away from passing, INP failures are usually structural: long JavaScript tasks blocking the main thread, oversized client bundles, third-party scripts firing on every click. It's not a metric a compression plugin fixes.
Where INP Actually Breaks
- Client components hydrating more state than the interaction actually needs
- Third-party analytics and chat widgets blocking the main thread on click
- Event handlers doing synchronous work that belongs in a worker or a debounce
- Large client bundles shipped to simple, mostly-static pages that never needed them
The Metric Coming Next: Visual Stability Index
Google is now collecting a new signal through CrUX called the Visual Stability Index (VSI), which tracks visual stability across an entire session, not just the layout shifts already counted by CLS, but instability introduced during scroll too. It isn't a hard ranking factor yet. Based on the rollout pattern of every Core Web Vitals metric before it, it's reasonable to expect that to change within 12 to 18 months. Sites that wait for it to become mandatory before addressing scroll-triggered layout shifts will be fixing under deadline what they could have fixed on a normal sprint.
What We're Doing About It
This is architecture work, not a plugin install. We build on Next.js specifically because server-side rendering and static generation put LCP within reach by default, and we treat INP the way we treat any latency budget in a backend system: profile first, then cut the main-thread work that's actually blocking, not the work that's easiest to blame. If your last performance audit was a Lighthouse score from last year, it isn't measuring what Google is scoring you on today, and it definitely isn't measuring what it will be scoring you on twelve months from now.