Most of the web is too slow to pass, and the bar is a business one. In 2025, only 48 percent of mobile sites and 56 percent of desktop sites met all three of the Google Core Web Vitals, the field measurements Google uses to score real-user experience and feed into search ranking. That means more than half of mobile sites fail, after years of industry attention. Two things make 2026 the year to fix it: responsiveness became a ranking metric of its own, and AI is now writing a large share of the frontend, in a way that makes these scores harder to pass, not easier.

What Core Web Vitals actually measure

Core Web Vitals are three field metrics, measured on real visitors rather than in a lab. Largest Contentful Paint (LCP) is how long the main content takes to appear; good is 2.5 seconds or less. Interaction to Next Paint (INP) is how quickly the page responds when a user taps or types; good is 200 milliseconds or less. Cumulative Layout Shift (CLS) is how much the page jumps around as it loads; good is 0.1 or less. A site passes only when the large majority of real visits clear all three. In March 2024, INP replaced an older metric and became the responsiveness Core Web Vital, which matters because interactivity is exactly where heavy modern pages struggle.

Most sites fail, and it is barely improving

The pass rate is low and moving slowly. Mobile Core Web Vitals climbed from 36 percent of sites in 2023 to 44 percent in 2024 and 48 percent in 2025, a few points a year despite a decade of tooling. Desktop sits higher at 56 percent and has been nearly flat. The pattern is telling: the problem is not that teams do not know performance matters. It is that pages keep getting heavier faster than teams optimize them, and the heaviest part is the code the browser has to run.

Why AI-built frontends make it harder

AI now generates a large and growing share of frontend code, and generated code optimizes for looking right, not for running light. It tends to ship more JavaScript, the code the browser downloads and executes to make a page interactive, and more third-party components than a human would hand-tune. More JavaScript is precisely what hurts INP, because every tap has to wait for the browser to finish its work, and it inflates LCP when the main content sits behind scripts. So the same shift that lets teams ship interfaces faster also pushes the numbers that decide ranking and revenue in the wrong direction, unless someone is watching them.

Why the score is a revenue number

Performance is not a vanity metric. Google research found that as mobile page load goes from one second to three, the probability that a visitor leaves rises 32 percent, and the effect compounds past three seconds. Slower pages lose the visit before the content ever converts, and because Core Web Vitals feed search ranking, a failing site is also harder to find in the first place. For a site that carries pipeline or revenue, a low pass rate is a direct tax on both traffic and conversion, paid quietly on every visit.

What to measure, starting now

Three moves turn this from a mystery into a managed number. First, measure field data, the scores from real visitors, not just a lab tool on your laptop, because real devices and networks are where sites actually fail. Second, set a performance budget: a hard ceiling on JavaScript weight and a target for each Core Web Vital, enforced on every change so a heavy AI-generated component fails the build instead of the field. Third, watch INP specifically, because it is the metric most sensitive to the script-heavy pages AI tends to produce, and the one teams most often miss. You cannot improve a number you do not look at, and most teams are not looking at this one.

Key Takeaways

  • Only 48 percent of mobile sites and 56 percent of desktop sites passed all three Core Web Vitals in 2025, so more than half of mobile sites fail.
  • The three metrics are LCP (loading), INP (responsiveness, a Core Web Vital since March 2024), and CLS (visual stability), all measured on real visitors.
  • AI-generated frontends ship more JavaScript, which hurts INP and LCP, so faster shipping quietly pushes the scores the wrong way.
  • The score is a revenue number: as load goes from one to three seconds the probability a visitor leaves rises 32 percent, and Core Web Vitals feed search ranking.
  • Measure field data, set a JavaScript and Core Web Vitals budget enforced on every change, and watch INP most closely.

Sources

  1. HTTP Archive, "The Web Almanac: Performance," 2025. Link.
  2. web.dev, "Interaction to Next Paint becomes a Core Web Vital on March 12," 2024. Link.
  3. Think with Google, "New Industry Benchmarks for Mobile Page Speed," 2018. Link.

Next Steps

The question for your team is not whether Core Web Vitals matter. It is whether anyone is holding your AI-generated frontend to them before it ships. Stable Solutions builds performance into the frontend as an R and D partner: field measurement, a Core Web Vitals and JavaScript budget in the pipeline, and INP profiling so the interface passes rather than just renders. Explore our App and Web Development work or contact our team to put a number on where your site stands today.