React — Core Web Vitals
The distribution, not just the median
Each bar is the good / needs-work / poor split for React. The marks where the whole fleet sits, so you can see who is ahead and who is behind.
What moves Core Web Vitals on React & Next.js
Field data findings across React & Next.js sites from our metric deep dives.
19.6% of React sites have loading="lazy" on their hero component.
Next.js <Image> sets loading="lazy" by default unless priority or preload is configured. Across React sites, 83% with default loading pass LCP compared with 87% with loading="lazy".
React sites with an HTML <img> hero pass LCP 18 percentage points more often than sites with CSS background heroes.
Across React sites, 84.1% include the LCP image in initial HTML (82% good LCP, 1.5s median) compared with 13.7% that reference the hero image in CSS (64% good LCP, 2.0s median).
React sites with all fonts on swap pass LCP 21 points more often than mixed-font sites.
Across React sites, 71.3% of @font-face rules use swap. Sites with all fonts on swap pass LCP at 90% (1.2s median) compared with 69% (1.9s median) on mixed-rule sites.
68.3% of React sites serve HTML over HTTP/3 (86% good LCP vs 69% on HTTP/2).
Across React sites, 68.3% serve main HTML over HTTP/3 (86% good LCP, 1.3s median) compared with 23.9% on HTTP/2 (69% good LCP, 1.9s median).
React sites using scheduler.yield() pass INP 2 points more often than sites using requestIdleCallback().
Across React sites, 62.9% call scheduler.yield() (100% good INP, 39ms median) compared with 22.9% calling requestIdleCallback() (98% good INP, 57ms median).
60.3% of React sites serve an AVIF hero image (1.2s median LCP).
Next.js Image Optimization supports image/avif and image/webp output formats. Across React sites, 60.3% serve AVIF hero images (92% good LCP, 1.2s median), 13.5% serve WebP, and 25% serve JPEG or PNG.
60.1% of React sites emit eager speculation rules and 16.8% emit no inline tag.
Next.js App Router provides route prefetching on <Link> components in JavaScript. Across React sites in the dataset, 60.1% carry inline eager speculation rules, 22.2% carry conservative rules, and 16.8% have no inline speculation rules tag.
35.3% of React & Next.js sites set fetchpriority="high" on the LCP element (88% vs 76% good LCP).
Next.js <Image> supports the fetchPriority="high" and priority props. Across React sites, 35.3% set fetchpriority="high" on the LCP element (88% good LCP, 1.3s median) compared with 60.9% with no hint (76% good LCP, 1.6s median, 12 percentage points apart).
React & Next.js sites with an <img> LCP element pass LCP 19 percentage points more often than sites with CSS background-image LCP.
Across React sites, 60.3% paint LCP from an <img> tag (84% good LCP, 1.4s median), 23.0% paint from text blocks (79% good LCP), and 14.6% paint from CSS background-image styles (65% good LCP, 1.9s median).
72.1% of React & Next.js sites load LCP media from external image origins (88% good LCP).
Next.js <Image> supports custom loader functions for external image CDNs. Across React sites, 72.1% load the LCP image from an external origin (88% good LCP, 1.3s median) and 27.9% load it from the application origin (70% good LCP, 1.8s median).
Among React & Next.js sites with speculation rules, 77.2% use prefetch and 5.7% use prerender_until_script.
Across React sites in the dataset, 77.2% emit prefetch rules (67% CWV pass rate), 5.7% emit prerender_until_script (92% CWV pass rate), and 16.8% emit no inline speculation rules tag.
React & Next.js sites on Brotli (70.6%) pass LCP 17 percentage points more often than sites on gzip (17.9%).
Across React sites, 70.6% serve Brotli-compressed HTML (86% good LCP, 1.4s median) compared with 17.9% serving gzip-compressed HTML (69% good LCP, 1.9s median).
React sites serving HTML from Fastly (37.1%) pass LCP at 91% vs 73% on direct origins.
Across React sites, 37.1% serve main HTML from Fastly edge nodes (91% good LCP, 1.2s median) compared with 46.9% serving HTML from direct origins (73% good LCP, 1.7s median, 18 points apart).
React & Next.js sites record a 63% all-three CWV pass rate across 7,993 sites.
Across 7,993 audited React & Next.js sites, 63% pass all three Core Web Vitals (LCP, INP, and CLS) simultaneously.
React & Next.js sites without CSS @import pass LCP 13 percentage points more often than sites with @import.
Across React sites, 96.0% ship stylesheets without runtime @import rules (80% good LCP, 1.5s median) compared with 4.0% that include a runtime CSS @import (67% good LCP, 1.8s median).
57.6% of React & Next.js sites ship at least one uncompressed JSON, SVG, or script response.
Across React sites, 42.4% compress every text asset (80% good LCP, 1.5s median) and 57.6% include at least one uncompressed text response (78% good LCP, 1.6s median).
99.6% of React & Next.js sites include Last-Modified headers on static and CDN responses.
Across React & Next.js deployments, 99.6% emit Last-Modified timestamps on static chunks or media assets (79% good LCP, 1.5s median).
33.1% of React sites lazy-load at least one above-the-fold image.
Next.js <Image> components default to loading="lazy" unless priority or preload is passed. Across React sites, 96.6% render an eager image in the viewport and 33.1% have a visible viewport image with loading="lazy".
React sites using <picture> (29.4%) pass LCP at 86%, 8 points above sites with fixed images.
Next.js <Image> generates responsive srcset candidate URLs. Across React sites, 84.9% use srcset on <img> (79% good LCP) and 29.4% use <picture> (86% good LCP).
React sites with HTML width/height (88.4%) pass CLS 18 points more often than sites with CSS-only aspect-ratio (5.1%).
Next.js <Image> requires width and height props unless fill is set. Across React sites, 88.4% include HTML width and height attributes (82% good CLS) and 5.1% use CSS aspect-ratio (64% good CLS).
Iframe-using React sites with sized iframes (62.5%) pass CLS 6 points more often than sites with unsized iframes.
Across 5,148 iframe-using React sites, 62.5% declare explicit width and height on their embeds (79% good CLS) compared with 70.4% with unsized iframes (73% good CLS).
Hint-using React sites include scroll-driven CSS on 77.3% and will-change on 71.3% (99% good INP).
Across 8,797 hint-using React sites, 77.3% include scroll-driven animation rules (99% good INP), 71.3% use will-change, and 53.5% use aspect-ratio.
Among shifting React sites, unsized_media sites pass field CLS at 46% and displaced sites at 72%.
Across 3,771 React sites that record a lab layout shift, 76.9% record box growth (66% good CLS), 35.4% record sibling displacement (72% good CLS), and 2.4% record unsized media shifts (46% good CLS).
React sites reach 98.8% HTTP/2 and 92.1% HTTP/3 subresource adoption (79% good LCP).
Across 9,282 React sites, 98.8% serve subresources over HTTP/2 and 92.1% serve subresources over HTTP/3 (79% good LCP).
70.8% of React & Next.js sites serve Cache-Control: immutable assets (86% good LCP).
Across React sites, 70.8% emit Cache-Control: immutable headers on static chunks (86% good LCP) compared with 75% on short-TTL sites.
27.7% of React sites also load jQuery (50% CWV pass vs 63% across all React sites).
Across 9,282 React sites, 27.7% load jQuery alongside React and pass all three Core Web Vitals at 50% (13 points below the 63% React cohort pass rate).
React sites with analytics use GA4 on 96.6% (99% good INP) and Matomo on 5.2% (99% good INP).
Across 4,068 analytics-using React sites, 96.6% load Google Analytics 4 (99% good INP), 71.0% include UA tags, and 5.2% run Matomo Analytics (99% good INP).
Ad-supported React sites pass all three CWV at 54% with AdSense and 44% with GPT.
Across 894 React sites running display ads, 93.5% load Google AdSense (54% passing all three CWV) and 8.9% load Google Publisher Tag (44%).
On the 597 WordPress + React sites using caching plugins, LiteSpeed Cache (26.1%) passes all three CWV at 43% and WP Rocket (34.7%) at 62%.
Across 597 React sites with a detected WordPress optimization plugin, 34.7% use WP Rocket (62% passing all three CWV), 26.1% use LiteSpeed Cache (43%), and 20.8% use Autoptimize (53%).
React sites with RUM (64.0%) pass all three CWV at 81% vs 57% on sites with tag managers (39.4%).
Across 9,280 React sites, 89.8% use a CDN (64% passing all three CWV), 64.0% include Real User Monitoring (81% CWV pass), and 39.4% load tag managers (57% CWV pass).
83% of React & Next.js sites have a DOM depth above 20 levels (40.3% at 32+ levels).
Across React sites, 42.6% have a maximum DOM depth of 21 to 31 levels (99% good INP) and 40.3% have 32+ levels (99% good INP).
67.8% of React & Next.js sites ship over 75 KB of HTML (84% good LCP).
Across React sites, 67.8% transfer over 75 KB of initial HTML (84% good LCP) and 16.8% transfer 30 to 75 KB (66% good LCP).
React & Next.js sites with 4+ iframes (8.1%) pass field CLS at 69% vs 84% with 0 iframes.
Across 9,282 React sites, 44.5% use 0 iframes (84% good CLS), 25.7% use 1 iframe (84%), and 8.1% embed 4 or more iframes (69% good CLS).
58.3% of React & Next.js sites record 0 lab layout shifts (91% good CLS vs 62% at 3 to 5 shifts).
Across 9,282 React sites, 58.3% record 0 lab shifts (91% good CLS), 26.0% record 1 to 2 shifts (79%), 9.2% record 3 to 5 shifts (62% good CLS), and 6.4% record 6+ shifts (54% good CLS).
Shifting React & Next.js sites with a largest shift over 0.25 (15.3%) pass field CLS at 51% vs 85% under 0.01 (34 points apart).
Across 3,867 shifting React sites, 36.3% have a largest shift under 0.01 (85% good CLS) and 15.3% exceed 0.25 (51% good CLS).
React & Next.js sites connecting to 15+ domains (16.6%) pass all three CWV at 57% vs 69% at 8 to 14 domains.
Across 9,282 React sites, 61.3% connect to 8 to 14 domains (69% passing all three CWV) and 16.6% connect to 15 or more domains (57%, 12 percentage points apart).
React & Next.js sites caching under 20% of resources (9.3%) pass LCP at 69% vs 81% at 80%+ (12 points apart).
Across React sites, 58.3% cache 80%+ of assets (81% good LCP), 29.0% cache 50% to 79% (80%), and 9.3% cache under 20% (69% good LCP).
React & Next.js sites with 9+ third-party providers (7.7%) pass all three CWV at 49% vs 79% at 0 to 1 providers (30 points apart).
Across 9,282 React sites, 43.2% load 0 to 1 external providers (79% passing all three CWV), 23.0% load 2 to 4 (68%), 26.0% load 5 to 8 (60%), and 7.7% load 9+ providers (49%).
React & Next.js sites serving AVIF (63.5%) pass LCP at 90% vs 74% on sites serving PNG (54.4%).
Next.js <Image> supports image/avif and image/webp output formats. Across 9,274 React sites, 63.5% serve AVIF images (90% good LCP, 1.2s median), 54.4% serve PNGs (74% good LCP, 1.7s median), and 44.2% serve JPEGs (74% good LCP).
93.9% of React & Next.js sites serve images from external image origins (79% good LCP vs 72% self-hosted).
Next.js <Image> supports custom loader functions for external media CDNs. Across 9,274 React sites, 93.9% serve images from external origins (79% good LCP, 1.5s median) and 32.3% serve images from the primary origin (72% good LCP, 1.8s median).
React & Next.js sites with synchronous blocking scripts (30.0%) pass LCP at 69% vs 80% with defer.
Across React sites, 89.2% load scripts with defer (80% good LCP), 96.8% use async (79% good LCP), and 30.0% include synchronous blocking scripts in <head> (69% good LCP, 1.9s median, 11 points apart).
94.8% of React & Next.js sites load cross-origin scripts (99% good INP, 48ms median).
Across 9,282 React sites, 100.0% run first-party bundles (99% good INP) and 94.8% also execute cross-origin scripts (99% good INP, 48ms median).
99.6% of React & Next.js sites declare scripts in HTML and 95.1% also load script-initiated chunks (99% good INP).
Across 9,282 React sites, 99.6% use parser-discovered <script> tags (99% good INP), 95.1% load script-initiated chunks (99% good INP), and 71.8% use other initiators (99% good INP).
React & Next.js sites with <body> stylesheets (14.1%) pass LCP 16 points less often than sites with inline CSS (63% vs 79%).
Across 9,281 React sites, 99.3% use inline <style> blocks (79% good LCP, 1.5s median), 56.0% load blocking <head> CSS (74%), and 14.1% include <link rel="stylesheet"> tags inside <body> (63% good LCP, 2.1s median).
React & Next.js sites loading third-party stylesheets (44.8%) pass LCP 6 points less often than sites on first-party CSS (73% vs 79%).
Across 9,281 React sites, 99.9% serve first-party CSS (79% good LCP, 1.5s median) and 44.8% load external third-party stylesheets (73% good LCP, 1.7s median).
99.3% of React & Next.js sites use inline styles (79% good LCP), while 37.4% load external <link> CSS (70%).
Across 9,281 React sites, 99.3% emit inline <style> tags (79% good LCP, 1.5s median), 36.2% load CSS from JS chunks (76% good LCP), and 37.4% load parser <link rel="stylesheet"> tags (70% good LCP, 1.8s median).
React & Next.js sites loading TTF fonts (7.3%) pass LCP 18 points less often than sites on WOFF2 (61% vs 79%).
Across 9,147 font-using React sites, 97.1% serve WOFF2 (79% good LCP, 1.5s median), 13.8% serve WOFF (68% good LCP), and 7.3% serve TTF (61% good LCP, 2.1s median).
React & Next.js sites loading external Google Fonts (18.9%) pass LCP at 65% vs 84% on dedicated CDNs.
Across 9,147 font-using React sites, 77.3% serve fonts from external CDNs (84% good LCP, 1.4s median) and 18.9% load Google Fonts at runtime (65% good LCP, 2.0s median, 19 percentage points apart).
85.6% of media-using React & Next.js sites stream from external CDNs (86% good LCP vs 69% self-hosted).
Across 1,803 React sites with <video> or <audio> tags, 85.6% serve media from external CDNs (86% good LCP, 1.3s median) and 15.9% serve media from their own origin (69% good LCP, 1.9s median, 17 points apart).
60.0% of LoAF React & Next.js sites record long frames in event listeners (99% good INP).
Across 5,701 React sites recording Long Animation Frames, 60.0% record long frames in DOM event listeners (99% good INP, 47ms median), 44.3% during script evaluation (98% good INP), and 18.6% in timer callbacks (99%).
91.6% of React & Next.js sites execute fetch() requests during load (79% good LCP, 1.5s median).
Across 9,282 React sites, 100.0% load JS bundles, 99.9% load images, 98.5% load web fonts (79% good LCP), and 91.6% fire fetch() requests (79% good LCP).
91.1% of React & Next.js sites serve Brotli-compressed subresources (79% good LCP, 1.5s median).
Across 9,282 React sites, 91.1% serve Brotli-compressed bundles (79% good LCP), 97.7% use Gzip (79% good LCP), and 41.3% include Zstandard-compressed scripts (74% good LCP).
74.6% of React & Next.js sites load subresources from CloudFront (82% good LCP) and 68.1% from Fastly (85%).
Across 9,282 React sites, 74.6% load subresources from Amazon CloudFront (82% good LCP, 1.4s median), 68.1% from Fastly (85% good LCP, 1.3s median), and 36.0% from Cloudflare (74% good LCP).
React apps exceeding 3 MB of images pass LCP 19 points less often than apps under 350 KB.
Across React sites, 38% transfer under 350 KB of images (86% good LCP, 1.3s median) compared with 14% transferring over 3 MB (67% good LCP, 1.9s median).
React sites under 40 KB of external CSS pass LCP 30 points more often than sites over 160 KB.
Across React sites, 55% ship under 40 KB of external CSS (88% good LCP, 1.3s median) and 22% ship over 160 KB (58% good LCP, 2.2s median).
React sites with over 600 KB of unused CSS pass LCP 28 points less often than sites under 140 KB.
Across React sites, 28% have under 140 KB of unused CSS (93% good LCP, 1.2s median) and 22% have over 600 KB of unused styles (65% good LCP, 2.0s median).
React sites loading 7+ font files pass LCP 12 points less often than sites with 1 to 3 fonts.
Across React sites, 27% load 1 to 3 font files (83% good LCP), 42% load 4 to 6 font files, and 30% load 7 or more font files (71% good LCP, 1.7s median).
81% of React apps load no audio or video streams on initial render.
Across React sites, 81% load no audio or video requests during initial navigation (78% LCP pass rate, 1.6s median) and 12% load under 500 KB (81%).
63% of React sites emit preload hints (88% good LCP vs 65% on dns-prefetch only).
Across React sites, 63% emit preload hints (88% good LCP, 1.3s median) compared with 9% emitting dns-prefetch only (65% good LCP, 2.1s median).
React sites with 6+ cookies pass LCP 24 points less often than sites with 3 to 5 cookies.
Across React sites, 43% set 3 to 5 cookies (89% good LCP, 1.3s median) and 20% set 6 or more cookies (65% good LCP, 2.0s median).
React sites painting under 300ms in the lab pass field LCP 28 points more often than sites over 1,500ms.
Across React sites, 46% paint their hero element in under 300ms in the lab (90% field LCP pass rate, 1.2s median) compared with 26% taking over 1,500ms (62% pass rate, 2.1s median).
React sites reaching DOMComplete under 350ms pass field LCP 32 points more often than sites over 1,500ms.
Across React sites, 48% reach DOMComplete in under 350ms (90% field LCP pass rate, 1.2s median) and 18% take over 1,500ms (58% pass rate, 2.2s median).
TTFB is the largest LCP sub-part on 57% of React sites (71% good LCP vs 88% when load delay leads, 17 points apart).
Across React sites, TTFB is the largest LCP sub-part on 57% (71% LCP pass rate, 1.8s median) and load delay is the largest sub-part on 38% (88% pass rate, 1.3s median).
React hero images over 300 KB pass LCP 18 points less often than hero images under 40 KB.
Across React sites, 49% serve an LCP image under 40 KB (87% LCP pass rate, 1.4s median), 29% serve 40 to 100 KB (79%), and 9% exceed 300 KB (69% pass rate, 1.9s median).
React sites exceeding 5 MB of total transfer pass LCP 27 points less often than sites under 1.2 MB.
Across React sites, 14% transfer under 1.2 MB (93% LCP pass rate, 1.0s median), 42% transfer 1.2 to 2.5 MB (89%), and 20% exceed 5 MB (66% pass rate, 1.9s median).
framework standings
210,902 crawled sites (69,745 with desktop Chrome field data), representing millions of real page loads, October 2026 edition. How we measured.