Website loading speed remains uneven across the web. In 2024, 43% of mobile websites and 54% of desktop websites achieved good Core Web Vitals when INP was used as the responsiveness metric. The same year, a median desktop page weighed 2,652 KB, while a median mobile page weighed 2,311 KB. These figures show why speed depends on more than a single load-time number: responsiveness, page weight, media, delivery, and user outcomes all matter.
Contents
- Core Web Vitals benchmarks
- The latest loading-speed metrics
- Page weight and JavaScript
- Images and fonts
- HTTP protocols and delivery
- Engagement and conversions
Core Web Vitals benchmarks
Core Web Vitals combine loading performance, responsiveness, and visual stability. The 2024 Web Almanac results show a persistent gap between mobile and desktop experiences. With INP used as the responsiveness metric, 43% of mobile websites achieved good Core Web Vitals in 2024, compared with 54% of desktop websites. These are broad website-level results from the Web Almanac performance data, rather than a guarantee for every page or visitor. (Performance, The Web Almanac 2024)
Popularity did not eliminate the gap. Among the 1,000 most popular websites, 40% achieved good Core Web Vitals on mobile and 54% did so on desktop. Across the top 10,000 websites, the figures were 33% on mobile and 46% on desktop. Across the top 100,000 websites, 31% of mobile websites and 43% of desktop websites achieved good results.
| Website group | Mobile with good Core Web Vitals | Desktop with good Core Web Vitals |
|---|---|---|
| All websites | 43% | 54% |
| Top 1,000 | 40% | 54% |
| Top 10,000 | 33% | 46% |
| Top 100,000 | 31% | 43% |
Page type also matters. Home pages had a 38% good Core Web Vitals rate on mobile in 2024, while secondary pages reached 51%. A homepage may carry branding, navigation, promotional media, and other first-visit elements, whereas secondary pages can have a different content and resource mix. The figures describe observed page groups; they do not establish that page type alone caused the difference.
The latest loading-speed metrics
Time to First Byte (TTFB) indicates how quickly the browser begins receiving a response. In 2024, 42% of mobile websites had good TTFB, 40% needed improvement, and 19% were poor. The good share was 41% in 2023, 40% in 2022, and 39% in 2021. (Performance, The Web Almanac 2024)
First Contentful Paint (FCP) measures when the first piece of page content appears. In June 2024, 68% of desktop websites had good FCP, compared with 51% of mobile websites. In September 2023, the corresponding shares were 63% for desktop and 47% for mobile. The periods are different, so these comparisons should be read as reported snapshots rather than a controlled year-over-year experiment.
Largest Contentful Paint (LCP) provides a view of when the largest visible content element renders. In 2024, 72% of desktop websites had good LCP, 20% needed improvement, and 8% were poor. On mobile, 59% had good LCP, 27% needed improvement, and 14% were poor. For home pages, good LCP reached 63% on desktop and 53% on mobile; secondary pages reached 82% and 72%, respectively.
The Web Almanac also reported LCP subpart medians for July 2024. For good p75 LCP, the median times were 600 milliseconds of TTFB, 350 milliseconds of image load delay, 160 milliseconds of image load duration, and 230 milliseconds of render delay. For poor p75 LCP, the corresponding figures were 2,270, 1,290, 350, and 360 milliseconds. This breakdown shows that a slow LCP can reflect several stages, not only image transfer.
Implementation details can obscure the element that controls LCP. In 2024, 35% of mobile pages lacked a statically discoverable LCP element. Also, 16% used native or custom lazy-loading on an image-based LCP element: native loading=lazy appeared on 9.5% of mobile pages and custom lazy-loading on 6.7%. (Performance, The Web Almanac 2024)
Page weight and JavaScript
Page weight has continued to rise over the long term. In October 2024, the median page weighed 2,652 KB on desktop and 2,311 KB on mobile. From October 2023 to October 2024, median desktop page weight grew 8.6%, or 210 KB. Median mobile page weight grew 6.4%, or 140 KB. (Page Weight, The Web Almanac 2024)
The ten-year comparison was larger: median desktop page weight was 120% higher in October 2024 than in 2014, while median mobile page weight was 357% higher. These are historical comparisons reported by the Web Almanac, not an estimate of the speed impact for a particular site.
JavaScript and images accounted for substantial portions of the median homepage in 2024. A median desktop homepage served 613 KB of JavaScript and 1,054 KB of images. A median mobile homepage served 558 KB of JavaScript and 900 KB of images. At the 75th percentile, desktop pages delivered 1,172 KB of JavaScript and mobile pages delivered 1,103 KB. At the 90th percentile, those amounts rose to 1,834 KB and 1,732 KB.
Request counts add another dimension. The median mobile page made 22 JavaScript requests in 2024, while the 90th percentile made 68. The median desktop page made 23 JavaScript requests, while the 90th percentile made 70. A page can therefore be relatively close to the median in one measure and much heavier in another, depending on how scripts are divided among requests. (JavaScript, The Web Almanac 2024)
Images and fonts
Images remain a major part of page delivery. In 2024, the median mobile image weighed 12 KB. The 75th-percentile image weighed 56 KB, and the 90th-percentile image weighed 177 KB. Looking specifically at the largest image on a page, the median mobile page had a largest image weighing 135 KB, while the 90th-percentile largest image weighed 1,002 KB. (Media, The Web Almanac 2024)
Image format adoption changed over the two-year comparison ending in 2024. AVIF adoption on mobile pages increased 386%, and WebP adoption increased 34%. The dataset reports these as adoption changes; it does not provide a universal speed improvement attributable to either format.
Fonts were also widespread in the 2024 HTTP Archive crawl. Web fonts appeared on about 87% of websites, and Google Fonts appeared on 48% of websites in the mobile crawl. The font-display: swap value was used on 45% of mobile pages. These figures describe usage and configuration prevalence, not the time added by each font implementation. (Fonts, The Web Almanac 2024)
Compression choices show a similar split. Brotli compression was used on 37% of mobile pages in 2024, while gzip was used on 52%. The reported shares concern compression usage, so they should not be interpreted as a direct ranking of the resulting page speeds without additional measurements. (Markup, The Web Almanac 2024)
HTTP protocols and delivery
Protocol support was mixed across mobile home pages in 2024. The main document loaded over HTTP/1.1 on 21% of mobile home pages, HTTP/2 on 70%, and HTTP/3 on 9%. Across more than 1.3 billion observed requests, 85% used HTTP/2 or HTTP/3. The home-page protocol shares and request-level share describe different scopes and should not be combined as if they measured the same population. (HTTP, The Web Almanac 2024)
Content delivery networks were also common: 54% of more than 1.3 billion observed requests were served from a CDN in 2024. HTTP/3 support was announced through alt-svc on 26% of desktop home pages and 28% of mobile home pages. Announcement support is not the same as proof that every visitor used HTTP/3, so the figures are best read as indicators of deployment or availability.
Engagement and conversions
Performance can matter commercially, but the strength and context of an association matter. In Deloitte’s four-week study, a 0.1-second natural mobile-speed improvement was associated with an 8.4% increase in retail conversions and a 9.2% increase in retail average order value. In travel, the same improvement was associated with a 10.1% increase in conversions and a 1.9% increase in average order value. (Milliseconds Make Millions, Deloitte)
The study also associated a 0.1-second natural mobile-speed improvement with an 8.6% increase in luxury-brand page views per session and an 8.3% improvement in the lead-generation information-page bounce rate. These are reported associations from Deloitte’s study, not universal forecasts for every industry or site.
A separate Akamai online-retail analysis, based on one month of user data and published in 2017, reported that a 100-millisecond delay in website load time hurt conversion rates by 7%. This is a legacy finding from that analysis and is not independently reverified here. (Akamai Online Retail Performance Report: Milliseconds Are Critical)
Together, the statistics point to several measurable performance priorities: Core Web Vitals outcomes are lower on mobile than desktop, page weight has grown substantially, JavaScript and the largest images can be sizable, and delivery infrastructure varies widely. Business studies connect small mobile-speed changes with engagement and conversion outcomes, but their study designs, dates, industries, and measures should remain explicit when applying them to a particular website.