(Technical)

Core Web Vitals

The three metrics and their thresholds

Core Web Vitals are metrics Google uses to quantify real users’ page experience along three axes: loading, responsiveness, and stability. LCP covers loading performance, INP covers interaction responsiveness, and CLS covers visual stability. On March 12, 2024, INP officially replaced FID (First Input Delay), and FID was removed from Search Console the same day. web.dev announcement

MetricWhat it measures“Good” threshold
LCP (Largest Contentful Paint)Time to render the largest content on screen (usually the hero image or headline)2.5 seconds or less
INP (Interaction to Next Paint)Delay from a click, tap, or key press to the next screen update200 ms or less
CLS (Cumulative Layout Shift)How much the layout shifts unexpectedly during loading0.1 or less

The verdict is based on real-user data, not lab scores. web.dev advises that “a good threshold to measure is the 75th percentile of page loads, segmented across mobile and desktop devices.” web.dev In other words, at least 75% of visits must fall within the thresholds on all three metrics to be rated “good,” and mobile and desktop are assessed separately.

Impact on search rankings: where official guidance and real-world data diverge

How much Core Web Vitals affect rankings is where Google’s official documentation and industry measurements differ most. The bottom line: they are a ranking signal, but they don’t beat content relevance.

Google’s official positionPage experience is a signal Google’s ranking systems consider, but content relevance comes first, and page experience acts more like a tiebreaker between otherwise similar results. Google (official)
Expert measurement resultsIn an Ahrefs study covering data on 5.2 million pages, the author said there was no way to draw conclusions about a correlation between Core Web Vitals and rankings, because the ranking update overlapped with other changes (2022). Ahrefs (data)

The gap exists because they measure different things. Google describes a policy (“we use it as a signal”), while measurements observe that “the signal doesn’t show up clearly in actual ranking changes.” Google itself wrote in its page experience documentation that “trying to get a perfect score just for SEO reasons may not be the best use of your time.” Google (official) In practice, rather than tuning these metrics alone and expecting a ranking jump, treating them as fundamentals within the bigger picture of SEO fits both positions.

Benefits beyond rankings: speed and conversion rates

Even if their ranking weight is small, the stronger reason to take these metrics seriously is conversion. The “Milliseconds Make Millions” study, commissioned by Google and conducted by Deloitte and 55, analyzed more than 30 million sessions across 37 brand sites in Europe and the US. A 0.1-second improvement in mobile speed increased retail conversion rates by 8.4% and retail order value by 9.2% (2020). Deloitte (data) Even at the same ranking, a slow site loses visitors after they arrive.

Lab data vs. field data: the score illusion

The PageSpeed Insights screen is actually two reports. According to Google’s documentation, the “Discover what your real users are experiencing” section at the top is field data collected by CrUX (Chrome User Experience Report) over the previous 28 days, while the 0–100 performance score below is lab data simulated by Lighthouse. PageSpeed docs Because the measurement conditions differ, the two can disagree.

INP in particular can’t be fully captured by lab tests. web.dev explains that “some lab tools can’t report INP because they observe only page loading, without interactions.” web.dev Check field-data pass rates for the whole site in the Core Web Vitals report in Google Search Console; that data also comes from CrUX and is shown separately for mobile and desktop. GSC Help Center

Practical checkpoints

Each item below is based on web.dev’s official optimization guide for that metric.

  • LCP – web.dev explicitly says to “never lazy-load your LCP image,” and recommends setting fetchpriority="high" on the image expected to be the LCP element and reducing render-blocking CSS/JS. web.dev
  • CLS – web.dev recommends reserving space for images and videos with width/height (or CSS aspect-ratio) and reserving space for late-loading ads and embeds with min-height. web.dev
  • INP – web.dev advises you to “do as little work as possible in event callbacks,” breaking up long tasks and running only the logic needed for the next frame’s visual update first. web.dev
<!-- LCP image: declare size (CLS) + load with priority (LCP) - per web.dev -->
<img src="hero.webp" width="1200" height="800"
     fetchpriority="high" decoding="async" alt="Main visual">

<!-- Preload key web font: reduce layout shifts caused by font swapping -->
<link rel="preload" href="/en/fonts/pretendard-sub.woff2"
      as="font" type="font/woff2" crossorigin>

Common mistakes

  • Treating a Lighthouse score as a “pass” – The verdict uses CrUX field data. That data reflects a 28-day collection window, so even after you ship improvements, the numbers move slowly. PageSpeed docs
  • An optimization plugin lazy-loading even the LCP image – web.dev states that lazy-loading the LCP image “always” adds an unnecessary resource load delay and hurts LCP. Always exclude above-the-fold images. web.dev
  • Measuring desktop only – Both web.dev and the Search Console documentation say mobile and desktop are assessed separately. Passing on one alone is not a full pass.
  • Confusing performance metrics with visibility problems – If a page doesn’t appear in search at all, look for the cause in crawling and indexing.

References

Jul.2026By Studio JT

More from Technical

Like this term?
Give it a like!

(43)