Core Web Vitals and Google Rankings Explained
Exactly how Google uses Core Web Vitals data for ranking, where to find your real metrics, and how to fix failing scores before they cost you organic traffic.
Since June 2021, Core Web Vitals have been an official Google ranking factor. But there’s a lot of confusion about how they work in practice — which metrics matter, where the data comes from, and how much impact failing scores actually has.
For the broader history of speed as an SEO factor, see How Page Speed Affects SEO Rankings. Here’s a definitive breakdown of the Core Web Vitals signal specifically.
What Google Measures
Google uses field data from the Chrome User Experience Report (CrUX). This is real performance data collected from real Chrome users who visited your site with usage statistics sharing enabled.
Not your Lighthouse score. Not your PageSpeed Insights lab results. Google looks at actual user experiences on actual devices and connections.
The three metrics:
LCP (Largest Contentful Paint) — how quickly the main content loads for real users. CLS (Cumulative Layout Shift) — how stable the layout is for real users. INP (Interaction to Next Paint) — how responsive the page is to real user interactions.
How Google Evaluates Your Site
For each URL, Google collects CrUX data over a rolling 28-day window. It evaluates each metric at the 75th percentile — meaning 75% of your real page views must hit “Good” for the URL to pass that metric.
A URL passes Core Web Vitals assessment when:
- 75th percentile LCP < 2.5s ✓
- 75th percentile CLS < 0.1 ✓
- 75th percentile INP < 200ms ✓
All three must pass. Failing any one means the URL fails the overall assessment.
Google evaluates at the page group level — URLs with similar patterns are often grouped and evaluated together, not individually.
Where to See Your Actual Data
Google Search Console → Core Web Vitals This is the canonical source. It shows how many of your URLs are in Good, Needs Improvement, and Poor categories for both desktop and mobile. It groups similar URLs and shows which URL patterns are failing.
PageSpeed Insights Enter any URL to see its CrUX field data alongside lab scores. For URLs with enough traffic, you’ll see the Core Web Vitals assessment result.
CrUX API / BigQuery For developers who want raw data. Google makes CrUX data available for free via API and monthly dataset releases.
What Happens When You Fail
Failing Core Web Vitals doesn’t make your page disappear from search results. Google has been clear: Core Web Vitals is a ranking factor among hundreds, not a hard filter.
The documented impact is:
- Sites with Good Core Web Vitals have a small ranking advantage over equivalent sites with Poor scores
- The effect is most noticeable in competitive niches where content quality is similar across multiple results
- There’s no reported “penalty” beyond losing the small advantage
Anecdotally, many site owners have reported meaningful ranking improvements after passing Core Web Vitals — particularly for competitive queries. But the causation is difficult to isolate.
The Bigger Picture
The SEO benefit of faster page speed goes beyond the direct Core Web Vitals signal:
- Lower bounce rates → more time on site → better behavioral signals
- Higher pages-per-session → more content indexed → more ranking opportunities
- Better crawl efficiency → more of your site crawled → faster index updates
A slow site signals poor quality to users, which creates signals that influence rankings beyond the explicit Core Web Vitals factor.
Fixing Core Web Vitals for SEO
- Start with Search Console — find which URL groups are failing and which metric.
- Test failing pages on debugutils.dev — get waterfall, LCP element, and prioritized recommendations.
- Fix LCP first — it’s the metric most sites fail and the highest-weight metric.
- Fix CLS — image dimensions, font loading, and dynamic content injection.
- Fix INP — reduce JavaScript on the main thread, defer non-critical scripts.
- Wait 28 days — CrUX data rolls over 28 days. Your Search Console report will update as the fix propagates.
The 28-day lag is important — make sure your fix is solid before expecting to see improvement in the data.