182 Website Speed Statistics October 2026

Website speed statistics from 2025 and 2026 describe a web that passes its performance tests more often every year and still misses them on close to half of mobile origins. Chrome's September 2026 dataset covered 18,376,044 origins (Chrome UX Report release notes, Sep 2026). The HTTP Archive crawl, platform engineering teams and the companies that measured what a faster page earned them supply the rest.

The page covers pass rates for loading, responsiveness and layout stability, how those rates differ by CMS, framework, device and country, page weight, JavaScript and third-party code, CDN, compression and caching adoption, what faster pages earned in conversion and revenue, and what Google has said about speed in ranking. Each number links to the organisation that produced it and carries the date of its data. Vendor research is labelled, case-study results are company-reported, and none of the figures comes from another roundup.

Conversion effects of load time are also covered in conversion rate optimization statistics, and video weight is covered in video SEO statistics. For the wider picture, start with SEO statistics.

How many sites pass Core Web Vitals, and how fast is the typical page?

Pass rates have climbed for six years and recently flattened on mobile. Loading, responsiveness and layout stability each have a different weak spot.

What share of sites passes Core Web Vitals in Chrome's data?

Slightly more than half, with a recent wobble in responsiveness. Chrome publishes the figures monthly from real-user data.

• In the September 2026 release, 68.3% of origins had good LCP, 81.7% had good CLS and 85.4% had good INP (Chrome UX Report release notes, Sep 2026).
• Good INP peaked at 87.2% of origins in March 2026 (Chrome UX Report release notes, Mar 2026).
• Chrome's August 2026 notes call the continued regression of INP a cause for concern and say the team has no definitive reason for it (Chrome UX Report release notes, Aug 2026).
• The share of origins with good Core Web Vitals was 47.3% in April 2024 and 51.0% in September 2024 (Chrome UX Report release notes, Sep 2024).
• An experimental variant that counts the soft navigations used by single-page apps puts the share of origins with good Core Web Vitals at 59.5% (Chrome UX Report release notes, Sep 2026).
• Nearly 33% of websites passed in Ahrefs' analysis of 2021 Chrome data, up about 10% on the year before, vendor research (Ahrefs, Jan 2022).
• Only 5.21 million of 43.66 million pages in Ahrefs' Site Audit data, about 11.9%, had at least one Core Web Vitals metric, vendor research (Ahrefs, Jan 2022).

How quickly do servers respond and pages first paint?

Fewer than half of mobile sites get a good server response, and the figure moved by only a point or two in a year.

• 55% of desktop sites and 44% of mobile sites had a good Time to First Byte of 0.8 seconds or less, up from 54% and 42% in 2024 (HTTP Archive Web Almanac, Jan 2026).
• 17% of mobile sites and 12% of desktop sites had a poor Time to First Byte, above 1.8 seconds (HTTP Archive Web Almanac, Jan 2026).
• Good First Contentful Paint rose from 68% to 70% of desktop sites and from 51% to 55% of mobile sites in a year (HTTP Archive Web Almanac, Jan 2026).
• 13% of mobile sites and 7% of desktop sites had a poor LCP, above 4.0 seconds (HTTP Archive Web Almanac, Jan 2026).
• web.dev says a well-optimized page spends about 40% of its LCP time on Time to First Byte, under 10% on resource load delay, about 40% on resource load duration and under 10% on element render delay (web.dev, Mar 2025).
• fetchpriority="high" sits on 17.3% of mobile pages with an LCP image, up from 15% in 2024, while preload sits on 2.1% (HTTP Archive Web Almanac, Jan 2026).

How responsive and visually stable are pages?

Desktop responsiveness is close to solved. Mobile responsiveness and layout stability are where the failures remain.

• 97% of desktop sites have good INP (HTTP Archive Web Almanac, Jan 2026).
• On mobile, 80% of home pages have good INP against 69% of secondary pages, a reversal from 2024, when the two sat within a point (73% against 72%) (HTTP Archive Web Almanac, Jan 2026).
• Total Blocking Time on mobile reached 4,193 milliseconds at the 75th percentile and 7,555 at the 90th, against 336 and 802 on desktop (HTTP Archive Web Almanac, Jan 2026).
• 72% of desktop pages and 81% of mobile pages have good CLS, up from 62% on both in 2021 (HTTP Archive Web Almanac, Jan 2026).
• 62% of mobile pages and 65% of desktop pages leave at least one image without explicit dimensions, down from 66% and 69% in 2024 (HTTP Archive Web Almanac, Jan 2026).
• Chrome usage data shows that 1 in 10 navigations on desktop and 1 in 5 on mobile are back or forward navigations (web.dev, Sep 2026).

Where do the Core Web Vitals thresholds come from?

From field data and perception research, judged at the 75th percentile, so a quarter of visits can be worse than the reported value.

• Google's documentation asks for LCP within 2.5 seconds, INP under 200 milliseconds and CLS under 0.1 (Google Search Central, Dec 2025).
• A site is judged at the 75th percentile of page views, so a 75th percentile LCP of 2 seconds is good and one of 5 seconds is poor (web.dev, May 2025).
• When Chrome set the LCP threshold from April 2020 data, 42% of phone origins and 51% of desktop origins met 2.5 seconds (web.dev, Apr 2020).
• A 4-second "poor" threshold would have classified roughly 26% of phone origins as poor in that April 2020 data (web.dev, Apr 2020).
• INP replaced First Input Delay as a Core Web Vital on March 12, 2024 (web.dev, Jan 2024).
• Jakob Nielsen's response-time limits are 0.1 second for a system to feel instantaneous, 1 second for an uninterrupted flow of thought and 10 seconds for keeping the user's attention (Nielsen Norman Group, Jan 1993).

Which CMS, builder or framework gives the best Core Web Vitals?

Tightly managed builders sit at the top of both the Web Almanac and the Tech Report, and extensible CMSs and client-rendered frameworks sit lower. Platform defaults reach every site on the platform at once.

How do the biggest CMSs compare on a phone?

Pass rates, load metrics and page weight each point to a different leader, and no CMS is best at all of them.

• Duda led mobile pass rates at 85% and TYPO3 followed at 79%, while Wix reached 74%, Squarespace 69%, Drupal 61% and WordPress 45% (HTTP Archive Web Almanac, Jan 2026).
• On LCP, 94% of Duda sites, 89% of TYPO3 sites and 81% of Wix sites were good on mobile, against 53% of WordPress sites and 54% across the CMSs compared (HTTP Archive Web Almanac, Jan 2026).
• Wix raised good CLS from about 86% to 95% of mobile sites, while Weebly fell to 58% (HTTP Archive Web Almanac, Jan 2026).
• On INP, Squarespace led at 96% good on mobile, ahead of TYPO3 at 95% and Duda at 94%, and 1C-Bitrix and Weebly trailed at 71% (HTTP Archive Web Almanac, Jan 2026).
• Median Lighthouse mobile performance scores were 64 for Wix, 58 for Webflow, 57 for Duda and 52 for Shopify, against 41 for WordPress and 40 for Joomla (HTTP Archive Web Almanac, Jan 2026).
• Median mobile JavaScript was 1,634 KB on Wix and 1,615 KB on Shopify, against 638 KB on WordPress and 453 KB on Joomla (HTTP Archive Web Almanac, Jan 2026).

Which builders and frameworks pass most often in September 2026?

Sites on hosted builders and static-first frameworks pass most often. Sites on client-rendered frameworks pass least, and the share is correlation across sites that use each technology.

• 85.8% of Duda origins pass Core Web Vitals on mobile (HTTP Archive Tech Report, Duda, Sep 2026).
• 76.2% of Shopify origins pass on mobile, the same share as a year earlier (HTTP Archive Tech Report, Shopify, Sep 2026).
• 71.3% of Webflow origins pass on mobile, up from 64.0% in September 2025 (HTTP Archive Tech Report, Webflow, Sep 2026).
• 71.0% of Squarespace origins pass on mobile, up from 65.0% (HTTP Archive Tech Report, Squarespace, Sep 2026).
• 71.1% of Astro origins pass on mobile, up from 61.6% (HTTP Archive Tech Report, Astro, Sep 2026).
• 35.7% of Next.js origins pass on mobile, up from 28.1% a year earlier (HTTP Archive Tech Report, Next.js, Sep 2026).
• 39.3% of WooCommerce origins pass on mobile, and only 12.1% have a good Time to First Byte (HTTP Archive Tech Report, WooCommerce, Sep 2026).

What have platform teams done to raise pass rates?

Platform work reaches millions of sites without any change by the site owner, which is why defaults move the totals.

• The WordPress performance team reported that the mobile Core Web Vitals pass rate across all WordPress sites rose from 28.31% to 36.44% during 2023 (WordPress Core, Dec 2023).
• WordPress mobile LCP passing rates rose from 34.48% to 43.37% in 2023, and the team traced the overall gain mainly to LCP (WordPress Core, Dec 2023).
• In WordPress 6.3, sites that stopped lazy-loading their LCP image saw LCP passing rates rise 16% to 21% on mobile (WordPress Core, Sep 2023).
• Sites that enabled WordPress speculative loading improved their LCP passing rate by about 1.9% at the median (WordPress Core, Mar 2025).
• WordPress 6.3's lab benchmarks showed block themes loading 27% faster and classic themes 18% faster (WordPress Core, Sep 2023).
• In WordPress 6.9 benchmarks, eliminating all render-blocking stylesheets from the Twenty Twenty-Five theme improved LCP by about 36% on an emulated Fast 4G connection, and inlining all styles improved it by about 45% on Slow 3G (WordPress Core, Nov 2025).
• GoDaddy took the Core Web Vitals pass rate of its Websites + Marketing customer sites from around 40% to almost 70% by focusing on CLS (web.dev, GoDaddy, Jul 2022).
• Shopify's streaming change improved FCP and LCP for the median store by 2% to 6%, which Shopify says matters more than the Time to First Byte gain because those are the metrics buyers see (Shopify Web Performance, Sep 2026).
• Shopify's automatic Early Hints made First Contentful Paint 76 milliseconds faster and LCP 100 milliseconds faster at the 75th percentile for eligible page views (Shopify Web Performance, Aug 2026).
• Moving Shopify's speculation rules from conservative to moderate eagerness cut median desktop Time to First Byte by 285 milliseconds and LCP by 228 milliseconds, with about 14% more HTML requests as the cost (Shopify Web Performance, May 2026).
• Shopify's first platform-wide speculation rules configuration improved loading metrics by an average of 130 milliseconds on desktop and 180 on mobile (Shopify Web Performance, Aug 2025).
• Nuvemshop raised the share of its stores with healthy LCP from 57% to 96% in a year after changing how it prioritised images (web.dev, Nuvemshop, Jun 2026).

How do devices, connections and countries change page speed?

Pass rates move with the visitor's country and device at least as much as with the site's build. Mobile pass rates differ by a factor of ten between the best and worst countries.

How much worse are phones than desktops?

Worse on every loading metric except layout stability, where phones do better.

• 65.6% of mobile origins have good LCP, against 79.0% on desktop (HTTP Archive Tech Report, Sep 2026).
• 83.4% of mobile origins have good CLS, against 74.6% on desktop, the one metric where phones lead (HTTP Archive Tech Report, Sep 2026).
• 79.4% of mobile origins have good INP, against 97.6% on desktop (HTTP Archive Tech Report, Sep 2026).
• 57.6% of mobile origins have good First Contentful Paint, against 75.4% on desktop (HTTP Archive Tech Report, Sep 2026).
• 45.9% of mobile origins have good Time to First Byte, against 58.2% on desktop (HTTP Archive Tech Report, Sep 2026).
• In Lighthouse lab tests, the median desktop page reached LCP in 2.2 seconds and the median mobile page in 4.7 seconds (HTTP Archive Web Almanac, Jan 2026).
• 91% of US adults own a smartphone, and 16% are smartphone-only internet users with no home broadband subscription (Pew Research Center, Nov 2025).
• 78% of US adults subscribe to a home broadband service (Pew Research Center, Nov 2025).
• iOS devices generated 35% of mobile device traffic worldwide in 2025, and more than half of request traffic came from mobile devices in 117 countries and regions (Cloudflare, Dec 2025).
• Safari 26.2 started exposing LCP and INP through the Performance API, so iOS visitors who were invisible in these metrics now appear in real-user dashboards (Shopify Web Performance, Dec 2025).

Search behavior by device is covered in search engine usage statistics.

Which countries pass Core Web Vitals most often?

Rich countries with fast networks pass most often, and large developing markets pass least, with the biggest gaps on LCP and INP.

• 71.4% of mobile origins in Japan pass (HTTP Archive Tech Report, Japan, Sep 2026).
• 69.2% of mobile origins in Germany pass (HTTP Archive Tech Report, Germany, Sep 2026).
• 63.3% of mobile origins in the United States pass, against 65.4% on desktop (HTTP Archive Tech Report, United States, Sep 2026).
• 33.7% of mobile origins in Brazil pass, and only 55.3% have good INP (HTTP Archive Tech Report, Brazil, Sep 2026).
• 31.6% of mobile origins in India pass, against 50.9% on desktop (HTTP Archive Tech Report, India, Sep 2026).
• 29.4% of mobile origins in Indonesia pass (HTTP Archive Tech Report, Indonesia, Sep 2026).
• 22.0% of mobile origins in South Africa pass (HTTP Archive Tech Report, South Africa, Sep 2026).
• 19.8% of mobile origins in Pakistan pass, and only 13.0% have a good Time to First Byte (HTTP Archive Tech Report, Pakistan, Sep 2026).
• Ahrefs found almost no sites on 3G or slower connections passing, and put the United States 38th by pass rate at 41% in August 2021, vendor research (Ahrefs, Jan 2022).

How fast are connections around the world?

The best and worst national medians differ about thirtyfold in both the mobile and fixed rankings, and 5G coverage is concentrated in rich countries.

• The global median mobile download speed was 202.21 Mbps in August 2026, with 24.15 Mbps upload and 29 milliseconds of latency (Ookla Speedtest Global Index, Aug 2026).
• The global median fixed broadband download speed was 256.16 Mbps, with 162.05 Mbps upload and 17 milliseconds of latency (Ookla Speedtest Global Index, Aug 2026).
• For mobile downloads the United States ranked 10th at 283.68 Mbps and India 50th at 164.97 Mbps (Ookla Speedtest Global Index, Aug 2026).
• The United Arab Emirates led mobile downloads at 763.70 Mbps and Eswatini ranked last of 102 countries at 24.68 Mbps (Ookla Speedtest Global Index, Aug 2026).
• Singapore led fixed broadband at 652.88 Mbps and Afghanistan ranked last of 149 countries at 21.36 Mbps (Ookla Speedtest Global Index, Aug 2026).
• Cloudflare's speed tests put Spain and Hungary above 300 Mbps average download in 2025, and Iceland had the lowest average idle latency at 13 milliseconds (Cloudflare, Dec 2025).
• 5G networks covered an estimated 55% of the world's population in 2025, with 84% coverage in high-income countries and 4% in low-income countries (ITU, Nov 2025).
• A typical user in a high-income country generates nearly eight times more mobile data than one in a low-income country (ITU, Nov 2025).
• About 6 billion people were online in 2025, and 94% of people in high-income countries use the internet against 23% in low-income countries (ITU, Nov 2025).
• 2.2 billion people remained offline in 2025 (ITU, Nov 2025).

How heavy are pages, and does weight cost speed?

Pages have grown every year for a decade, and heavier home pages pass Core Web Vitals less often at every step up in weight.

How fast is the median page growing?

The median home page grew 7.8% in a year, and inner pages grew faster.

• The median mobile home page weighed 2,362 KB in July 2025, against 845 KB in July 2015, a 202.8% increase, while the desktop median rose 110.2% (HTTP Archive Web Almanac, Jan 2026).
• The median desktop home page reached 2.9 MB, up 7.3% from 2.7 MB in 2024 (HTTP Archive Web Almanac, Jan 2026).
• Median inner pages weighed 1.8 MB on mobile and 2 MB on desktop, and inner page weight grew 9.5% in a year (HTTP Archive Web Almanac, Jan 2026).

What are the bytes made of?

Images and JavaScript make up most of the weight, and HTML is a rounding error.

• The median mobile home page carries 911 KB of images, 632 KB of JavaScript, 122 KB of fonts, 77 KB of CSS and 22 KB of HTML (HTTP Archive Web Almanac, Jan 2026).
• The median desktop home page carries 1,058 KB of images and 697 KB of JavaScript (HTTP Archive Web Almanac, Jan 2026).
• Inner pages carry 354 KB of images on mobile, and the median home page carries 239% of the image bytes of an inner page on desktop (HTTP Archive Web Almanac, Jan 2026).

Does page weight predict Core Web Vitals?

Yes, with a caveat. The data covers only pages with URL-level Chrome data, which skews toward popular sites.

• 70% of desktop home pages under 1 MB pass Core Web Vitals, against 38% of those at 5 MB or more (HTTP Archive Web Almanac, Jan 2026).
• Inner pages hold up better: 79% of desktop inner pages under 1 MB pass, against 50% at 5 MB or more (HTTP Archive Web Almanac, Jan 2026).
• Home page LCP rises from 2.1 seconds on desktop and 2.6 seconds on mobile for pages under 1 MB to 3.3 and 3.8 seconds for pages of 5 MB or more (HTTP Archive Web Almanac, Jan 2026).
• 88% of mobile home pages under 1 MB have good CLS, against 71% of those at 5 MB or more (HTTP Archive Web Almanac, Jan 2026).

How are images and fonts delivered?

Legacy formats still carry most of the weight, and fonts are mostly self-hosted WOFF2.

• JPG is the format of 57% of LCP images and PNG of 26%, while WebP is at 11% and AVIF at 0.7% (HTTP Archive Web Almanac, Jan 2026).
• 22.5% of websites use WebP and 1.7% use AVIF, against 69.3% for JPEG (W3Techs, Oct 2026).
• WebP lossless images are 26% smaller than PNGs, and lossy WebP images are 25% to 34% smaller than comparable JPEGs (Google WebP documentation, Aug 2025).
• In Chrome tests, 97.5% of lazy-loaded images on 4G were fully loaded within 10 milliseconds of becoming visible, and 92.6% on slow 2G (web.dev, Aug 2024).
• Self-hosted fonts appear on over 71% of pages and Google Fonts on about 47% (HTTP Archive Web Almanac, Jan 2026).
• 50.1% of mobile sites set font-display: swap, 24.9% use block, and only 11.7% preload a font (HTTP Archive Web Almanac, Jan 2026).

How much JavaScript and third-party code do pages run?

Script bytes cost more than image bytes of the same size, because the browser has to parse and run them. Third-party code adds requests the site owner does not control.

How much JavaScript ships with a page, and how much goes unused?

Nearly every page uses it, and the median page carries hundreds of kilobytes it does not use on load.

• The median page makes 22 JavaScript requests on mobile and 23 on desktop (HTTP Archive Web Almanac, Jan 2026).
• The median page carries 251 KB of unused JavaScript on mobile and 280 KB on desktop, measured uncompressed (HTTP Archive Web Almanac, Jan 2026).
• jQuery runs on 65.3% of websites, Bootstrap on 14.8% and React on 6.1% (W3Techs, Oct 2026).
• V8's team measured JavaScript processing at 10% to 30% of the time spent in V8 during a typical page load on a high-end desktop machine (V8, Jun 2019).

How common are third-party scripts and requests?

The Web Almanac describes third parties as ubiquitous and increasingly concentrated in a few vendors.

• Scripts make up 24.8% of third-party requests, images 19.9% and other types 13.9% (HTTP Archive Web Almanac, Jan 2026).
• The top 10 third-party domains are dominated by Google services, and facebook.com is the only non-Google domain among them, on 21% of pages (HTTP Archive Web Almanac, Jan 2026).
• 77% of desktop home pages and 41% of mobile home pages could save weight by loading embeds behind a facade, according to Lighthouse (HTTP Archive Web Almanac, Jan 2026).
• Google Tag Manager runs on 45.1% of websites (W3Techs, Oct 2026).

What changes when third-party code is fixed?

Several vendors have published the before and after.

• Poptin cut Interaction to Next Paint at the 75th percentile from 280 to 120 milliseconds, a 57% reduction, across thousands of Shopify stores after shrinking a main-thread task of about 900 milliseconds to about 75 (Shopify Web Performance, Aug 2026).
• The share of origins passing INP when using PubTech's consent management platform rose from 13% to 55% on mobile after the vendor broke up long tasks (web.dev, PubTech, Feb 2024).
• Taboola used Long Animation Frames data to improve INP by up to 36% for publisher partner websites (web.dev, Taboola, Feb 2024).
• Google's ads team measured a 0.5% revenue lift for publishers and a 2% rise in early ad script loads after serving its ads script with stale-while-revalidate (web.dev, Google Ads, Mar 2020).
• 46.5% of origins using Google Tag Manager pass on mobile, a share that reflects everything else those sites ship as well as the tag (HTTP Archive Tech Report, Google Tag Manager, Sep 2026).
• 58.0% of origins using the Cookiebot consent tool pass on mobile (HTTP Archive Tech Report, Cookiebot, Sep 2026).

How widely are CDNs, compression, caching and newer protocols used?

CDNs and compression are widely adopted. Early Hints and speculation rules are rare outside the platforms that switch them on for everyone.

How many sites sit behind a CDN?

A quarter of websites use Cloudflare as a CDN, and CDN use varies sharply by request type.

• CDNs served 71% of third-party requests, 52% of subdomain requests and 35% of HTML requests on mobile (HTTP Archive Web Almanac, Jan 2026).
• Cloudflare served 58% of CDN-delivered HTML requests, Google 21%, CloudFront 7%, Fastly 5% and Vercel 2% (HTTP Archive Web Almanac, Jan 2026).
• 25% of all websites use Cloudflare as a CDN, up from 19.6% a year earlier and 3.9% ten years earlier (W3Techs, Sep 2026).
• Cloudflare is the web server of 31.9% of websites whose server is known, ahead of Nginx at 30.6%, Apache at 21.7% and LiteSpeed at 14.1% (W3Techs, Oct 2026).
• Cloudflare says its network sits within 50 milliseconds of 95% of the internet-connected population (Cloudflare, Sep 2024).

How common are compression and HTTP/3?

Compression is near universal, and the newest protocol is on four in ten sites.

• 93.3% of websites use compression (W3Techs, Oct 2026).
• HTTP/3 is in use on 40.9% of websites and HTTP/2 on 34.8% (W3Techs, Oct 2026).
• CDNs served mobile HTML over HTTP/3 29% of the time and HTTP/2 69%, while origin servers served none over HTTP/3 and 79% over HTTP/2 (HTTP Archive Web Almanac, Jan 2026).
• CDNs served 46% of mobile requests as Brotli, 42% as gzip and 12% as Zstandard, while origin servers served 39% as Brotli and 61% as gzip (HTTP Archive Web Almanac, Jan 2026).
• 72% of mobile home pages pass Lighthouse's text compression audit, and over 25% of pages across all devices and page types fail it (HTTP Archive Web Almanac, Jan 2026).
• Brotli made JavaScript files 15% smaller than gzip, HTML 18% smaller and CSS and font files 17% smaller for The Economic Times (web.dev, The Economic Times, Dec 2021).
• QuintoAndar found Brotli through its CDN cut JavaScript size 18% against gzip, and Brotli at build time cut it 24% (web.dev, QuintoAndar, Dec 2021).
• 15 countries and regions sent more than a third of their requests to Cloudflare over HTTP/3 in 2025, with Georgia highest at 38% (Cloudflare, Dec 2025).

Do caching, Early Hints and speculative loading deliver?

Early Hints and speculation rules deliver measurable gains and are rare outside platforms that switch them on for everyone.

• The median desktop home page could save 303 KB and the median mobile one 270 KB if resources with short cache lifetimes were served from cache (HTTP Archive Web Almanac, Jan 2026).
• Early Hints appear on barely over 6% of desktop pages in the top million sites, and on most other rank groups well below 5% (HTTP Archive Web Almanac, Jan 2026).
• Speculation Rules appear on 15% of home pages in the top million sites (HTTP Archive Web Almanac, Jan 2026).
• In Cloudflare's benchmark, Early Hints improved FCP by more than 9.47% and LCP by more than 6.03% for half of desktop pages (Cloudflare, Jun 2022).
• Cloudflare measured LCP improvements of 40% to 50% at the 75th percentile, around 45% on average, for sites with Speed Brain prefetching (Cloudflare, Sep 2024).
• About half of the hundreds of billions of HTML requests Cloudflare sees each day are cached (Cloudflare, Sep 2024).
• Over 8% of Chrome navigations rely on Speculation Rules (WordPress Core, Mar 2025).
• Ray-Ban used prerendering with Speculation Rules and improved LCP by 43% on both desktop and mobile (web.dev, Ray-Ban, Jan 2025).
• Monrif's prerendering and back/forward cache work cut desktop LCP by 433 milliseconds, 17.9%, from 2,419 (web.dev, Monrif, Dec 2025).

What does a faster page earn in conversion, engagement and revenue?

Companies that measured their own sites report gains in conversion, engagement and ad revenue after fixing LCP, INP and CLS. Each result comes from one site's own analytics, and the effect sizes vary widely.

How do load times and conversion move together in retail and travel?

Consistently in the same direction. The size of the effect depends on the site and the starting point.

• T-Mobile recorded a 20% reduction in user website complaints and a 34% reduction in slow-loading complaints after its LCP work (web.dev, T-Mobile, Mar 2025).
• Ray-Ban's prerendering raised product page conversion rates by 101.47% on mobile and 156.16% on desktop in A/B testing, and cut exit rates by about 13% on both (web.dev, Ray-Ban, Jan 2025).
• Zalando's own analysis found that shorter loading times went hand in hand with higher revenue per session at every step of the journey and on every device (Zalando Engineering, Jun 2018).
• Over 75% of Zalando's traffic came from mobile devices in 2018, and it cut overall loading time by 25% in five months (Zalando Engineering, Jun 2018).
• Shopify's director of performance engineering said that when a buyer's first page is 10% faster, conversion rises 7% on average (Cloudflare, Jun 2022).
• Farfetch found conversion rate falls on average 1.3% with each extra 100 milliseconds of LCP beyond the recommended point, and rises 2.8% for each second cut from Time to Interactive (web.dev, Farfetch, Jul 2022).
• Renault analysed over 10 million visits in 33 countries and found a 1-second LCP improvement could cut bounce rate by 14 percentage points and raise conversions by 13% (web.dev, Renault, Oct 2021).
• Swappie's relative mobile conversion rate went from 24% to 34% in three months, and mobile revenue rose 42% (web.dev, Swappie, Sep 2021).
• redBus raised sales by about 7% after improving INP (web.dev, redBus, May 2023).
• Tokopedia's performance work raised click-through rate 35% and conversion rate 8% (web.dev, Tokopedia, Oct 2020).

What do publishers and ad platforms gain?

Longer sessions and more ad views, which is where publishers see revenue.

• Yahoo! JAPAN News raised its mobile back/forward cache hit rate from 0.04% to 54.07% and saw a 9.0% rise in mobile ads revenue, against 0.6% on desktop (web.dev, Yahoo! JAPAN News, Jul 2023).
• Yahoo! JAPAN News cut CLS by 0.2, which raised page views per session 15%, session duration 13% and lowered bounce rate by 1.72 percentage points (web.dev, Yahoo! JAPAN News, Mar 2021).
• Netzwelt raised advertising revenue 18%, page views 27% and ad viewability above 75% while cutting bounce rate 50% (web.dev, Netzwelt, Jul 2021).
• The Economic Times cut LCP from 4.5 to 2.5 seconds and reduced bounce rate by 43% overall (web.dev, The Economic Times, Dec 2021).
• A later Economic Times INP project cut Total Blocking Time 30 times and INP nearly fourfold, with a 50% lower bounce rate and 43% more page views (web.dev, The Economic Times, May 2023).
• NDTV improved LCP 55% in a month, which correlated with a 50% reduction in bounce rate (web.dev, NDTV, Oct 2020).
• Terra's prefetching raised ads click-through rate 11% on mobile and 30% on desktop (web.dev, Terra, Sep 2023).
• Mail.ru raised the conversion rate of its core sections by more than 10% after improving CLS (web.dev, Mail.ru, Nov 2021).

What did fixing INP deliver?

Engagement and conversions, with INP covering every interaction on the page.

• QuintoAndar cut INP by 80%, took the share of pages with good INP from 42% to 78%, and reported a 36% year-over-year rise in conversions (web.dev, QuintoAndar, Jan 2025).
• Disney+ Hotstar cut INP on living-room devices by 61% and weekly card views per user rose from 111 to 226 (web.dev, Disney+ Hotstar, May 2024).
• Trendyol halved INP and saw a 1% uplift in click-through rate on search (web.dev, Trendyol, Dec 2023).
• Fotocasa's INP work contributed to a 27% increase in contact and phone lead ads (web.dev, Fotocasa, Oct 2025).
• Monrif's speculative loading and back/forward cache work improved engagement by 8.9% in some segments (web.dev, Monrif, Dec 2025).

Is page speed a Google ranking factor?

Yes, in a limited way, and Google's wording has stayed consistent since 2010. Speed breaks ties between similar pages and demotes the slowest ones, and relevance still wins first.

What has Google said about speed in ranking since 2010?

The same three things: speed is a signal, it is a small one, and it applies a single standard to every page.

• Google's 2010 announcement said site speed did not carry as much weight as the relevance of a page, and that the signal applied only to English searches on Google.com at launch (Google Search Central Blog, Apr 2010).
• Google announced in January 2018 that page speed would be a ranking factor for mobile searches from July 2018 (Google Search Central Blog, Jan 2018).
• The 2018 Speed Update affects only the pages that deliver the slowest experience and a small percentage of queries, applies the same standard to all pages whatever the technology, and a slow page may still rank highly with great, relevant content (Google Search Central Blog, Jan 2018).
• In May 2020 Google said that when multiple pages have similar content, page experience becomes much more important for visibility (Google Search Central Blog, May 2020).
• The page experience update began rolling out in mid-June 2021 and played its full role by the end of August 2021 (Google Search Central Blog, Apr 2021).
• The desktop rollout began in February 2022 and finished by the end of March 2022, using the same three Core Web Vitals (Google Search Central Blog, Nov 2021).
• In November 2020 Google said engagement with Lighthouse and PageSpeed Insights had risen by a median of 70% since the announcement (Google Search Central Blog, Nov 2020).
• Google retired Search Console's Mobile Usability report and the Mobile-Friendly Test from December 1, 2023, while keeping the Core Web Vitals and HTTPS reports (Google Search Central Blog, Apr 2023).

How does Google use Core Web Vitals in ranking today?

As one input among many, judged per page, with some site-wide assessment.

• Google says good scores in Search Console or third-party tools do not guarantee a top ranking, and that chasing a perfect score for SEO alone may not be the best use of time (Google Search Central, Sep 2026).
• Google's core ranking systems generally evaluate content on a page-specific basis, with some site-wide assessments (Google Search Central, Sep 2026).
• Google says it always seeks to show the most relevant content even when page experience is sub-par, and that a great page experience matters most when plenty of helpful content is available (Google Search Central, Sep 2026).
• PageSpeed Insights reports field data over a trailing 28-day collection period (Google PageSpeed Insights documentation, Oct 2024).
• Google's search results page was slow enough in 2017 that, for certain popular queries, 25% of the time it took more than three seconds to load, and Bing's first result arrived sooner 98% of the time (US District Court for D.C., Aug 2024).
• The court record says Google launched Project Folly to cut that latency, after finding its own results about 300 milliseconds slower than Bing's (US District Court for D.C., Aug 2024).

What do ranking studies find?

That speed does not separate pages on page one, which fits Google's own description of the signal.

• Backlinko compared Alexa domain load times for 1 million domains with rankings and found zero correlation with first-page positions, vendor research (Backlinko, Mar 2020).
• Ahrefs said that the timing of the rollout and other simultaneous updates made it impossible to isolate the effect of Core Web Vitals on rankings, and that sites that work on speed are more likely to work on SEO generally, vendor research (Ahrefs, Jan 2022).

Ranking factors beyond speed are covered in website SEO statistics.

Methodology and Sources

Every statistic on this page was traced to the organisation that produced it: the Chrome UX Report, the HTTP Archive (Web Almanac 2025 and the Core Web Vitals Technology Report), Google Search Central, web.dev, a US federal court record, W3Techs, Ookla, ITU, Pew Research Center, Cloudflare, Shopify, the WordPress Core performance team, Zalando, Nielsen Norman Group, V8, and Google's WebP documentation. The vendor studies are Ahrefs and Backlinko. The company results from web.dev are published by the companies themselves, measure one site each and are company-reported. Country and technology pass rates in the Tech Report are shares of origins that Chrome has enough data on, and they are correlations across sites using a technology, not effects of the technology. Each link was opened on 14 October 2026 to confirm the page loads, the number appears on it and the date is right, and the Tech Report figures were also read from the data behind the report. Where two figures from different chapters or dates disagree, the page uses the one stated for the date shown. We do not cite statistics roundups. The page is refreshed monthly, and the Chrome and Tech Report figures are replaced first.

Glossary

• Core Web Vitals: Google's three field metrics for loading (LCP), responsiveness (INP) and visual stability (CLS).
• LCP: Largest Contentful Paint, the time until the largest visible element has rendered.
• INP: Interaction to Next Paint, how quickly a page responds to clicks, taps and key presses.
• CLS: Cumulative Layout Shift, how much visible content moves unexpectedly.
• FCP: First Contentful Paint, the time until any content first appears.
• TTFB: Time to First Byte, the time until the first byte of the response arrives.
• TBT: Total Blocking Time, a lab measure of main-thread blocking that stands in for INP.
• CrUX: Chrome UX Report, the public dataset of real-user measurements from Chrome.
• Origin: a scheme, host and port, used as the unit of measurement in CrUX.
• 75th percentile: the value that 75% of page views meet or beat, used to judge Core Web Vitals.
• CDN: content delivery network.
• Early Hints: an HTTP status that lets a server tell the browser which resources to fetch before the page response is ready.
• Speculation Rules: a browser API for prefetching or prerendering pages the visitor is likely to open next.
• bfcache: the back/forward cache, which restores a page from memory on back and forward navigation.
• Brotli: a compression format that typically beats gzip on text files.

Sources

• Ahrefs: Core Web Vitals Data Study with CrUX and 5.2M Pages
• Backlinko: We Analyzed 11.8 Million Google Search Results
• Chrome UX Report: Release notes
• Cloudflare: Introducing Speed Brain
• Cloudflare: Early Hints performance
• Cloudflare: The 2025 Radar Year in Review
• Google PageSpeed Insights documentation: About PageSpeed Insights
• Google Search Central Blog: Using site speed in web search ranking
• Google Search Central Blog: Using page speed in mobile search ranking
• Google Search Central Blog: Evaluating page experience for a better web
• Google Search Central Blog: Timing for bringing page experience to Google Search
• Google Search Central Blog: More time, tools and details on the page experience update
• Google Search Central Blog: Timeline for bringing page experience ranking to desktop
• Google Search Central Blog: The role of page experience in creating helpful content
• Google Search Central: Understanding Core Web Vitals and Google search results
• Google Search Central: Understanding page experience in Google Search results
• Google WebP documentation
• HTTP Archive: Core Web Vitals Technology Report
• HTTP Archive: Web Almanac 2025, CDN
• HTTP Archive: Web Almanac 2025, CMS
• HTTP Archive: Web Almanac 2025, Fonts
• HTTP Archive: Web Almanac 2025, Page Weight
• HTTP Archive: Web Almanac 2025, Performance
• HTTP Archive: Web Almanac 2025, Third Parties
• ITU: Facts and Figures 2025 press release
• Nielsen Norman Group: Response Times, The 3 Important Limits
• Ookla: Speedtest Global Index
• Pew Research Center: Mobile Fact Sheet
• Pew Research Center: Internet and Broadband Fact Sheet
• Shopify Web Performance: Early Hints
• Shopify Web Performance: Liquid storefronts now start loading 29% faster
• Shopify Web Performance: Moderate speculation rules
• Shopify Web Performance: Poptin
• Shopify Web Performance: Safari LCP and INP
• Shopify Web Performance: Speculation Rules at Shopify
• US District Court for D.C.: United States v. Google LLC, Memorandum Opinion (Aug 2024)
• V8: The cost of JavaScript in 2019
• W3Techs: Image file formats
• W3Techs: JavaScript libraries
• W3Techs: Reverse proxy services
• W3Techs: Site elements
• W3Techs: Tag managers
• W3Techs: Web servers
• web.dev: Back/forward cache
• web.dev: Browser-level image lazy-loading
• web.dev: Defining the Core Web Vitals metrics thresholds
• web.dev: INP becomes a Core Web Vital on March 12
• web.dev: Optimize Largest Contentful Paint
• web.dev case study: Farfetch
• web.dev case study: Fotocasa
• web.dev case study: GoDaddy
• web.dev case study: Google ads and stale-while-revalidate
• web.dev case study: Disney+ Hotstar
• web.dev case study: Mail.ru
• web.dev case study: Monrif
• web.dev case study: NDTV
• web.dev case study: Netzwelt
• web.dev case study: Nuvemshop
• web.dev case study: PubTech
• web.dev case study: QuintoAndar
• web.dev case study: QuintoAndar INP
• web.dev case study: Ray-Ban
• web.dev case study: redBus
• web.dev case study: Renault
• web.dev case study: Swappie
• web.dev case study: T-Mobile
• web.dev case study: Taboola
• web.dev case study: Terra
• web.dev case study: The Economic Times
• web.dev case study: The Economic Times INP
• web.dev case study: Tokopedia
• web.dev case study: Trendyol
• web.dev case study: Yahoo! JAPAN News bfcache
• web.dev case study: Yahoo! JAPAN News CLS
• WordPress Core: WordPress performance impact on Core Web Vitals in 2023
• WordPress Core: WordPress 6.3 Core Web Vitals field analysis
• WordPress Core: WordPress 6.9 frontend performance field guide
• WordPress Core: Speculative loading in 6.8
• Zalando Engineering: Loading time matters