Page Speed Statistics for 2026: How Fast Real Websites Are, and What the Studies Say a Slow Second Costs

Sixteen page speed figures, each with a date and a publisher

Every figure below was opened on the publisher's own page in October 2026. Where the best evidence is older than 2024, the row says so, because a study run on 3G networks describes a different web. This page belongs to our marketing statistics library, whose front page quotes the 48% mobile pass rate as a headline; here it is the starting line.

The web's report card: three years of Core Web Vitals pass rates

The Web Almanac, published by the HTTP Archive, pairs its own crawl with the Chrome UX Report, Google's public dataset of what real Chrome users experienced. Its 2025 performance chapter, using July 2025 data, shows the mobile pass rate climbing from 36% in 2023 to 44% in 2024 and 48% in 2025. Desktop moved from 48% to 55% to 56%. Progress, then, at the pace of a municipal road project.

Thresholds come from Google's Core Web Vitals documentation, and INP replaced First Input Delay as the responsiveness metric on March 12, 2024. Two details in the Almanac deserve more attention than the headline. Popularity does not buy speed: the top 1,000 sites pass on mobile 51% of the time, the next 10,000 only 42% and the next 100,000 just 37%. And home pages fail more often than inner pages, 45% against 56% on mobile, so the page a business spends most on designing is the one most likely to keep a phone waiting.

Why LCP fails: the hero photo nobody resized

Largest Contentful Paint measures when the biggest thing on the screen appears, and on 76% of mobile pages that thing is an image. The Almanac's figures read like a list of avoidable mistakes. Roughly 16% to 17% of pages lazy-load their LCP image, telling the browser to fetch the most important picture last. Only about 17% of mobile pages with an LCP image mark it with fetchpriority="high". And 62% of mobile pages carry at least one image with no width or height set, the classic cause of layout shift.

• Pages tripled in a decade. The median mobile home page weighed 845 KB in July 2015 and 2.56 MB in July 2025, with 911 KB of images and 632 KB of JavaScript, per the Almanac's page weight chapter.
• Scripts got heavier still. Median Total Blocking Time on mobile reached 1,916 milliseconds in 2025, up 58% from 1,209 a year earlier.
• Modern formats remain rare. JPG served 57% of LCP images, WebP 11% and AVIF 0.7%.
• Servers start late. Fewer than half of mobile sites, 44%, had a good Time to First Byte, which delays everything after it.

None of these is exotic. Each is a setting, a file size or a hosting choice, which is why our guide to improving Core Web Vitals starts with the hero image, and image SEO covers formats and sizing. When the first byte is the problem, slow shared hosting is the usual suspect.

Pass rates by platform, and why the averages flatter some

The Almanac's 2025 CMS chapter, looking at mobile, puts Duda first with 85% of sites passing Core Web Vitals, Wix at 74% after a jump from 55% the year before, and WordPress among the lowest at 45%. WordPress runs roughly 64% of CMS-driven sites, so its average carries the most weight and the widest spread of quality.

A tuned WordPress site can pass comfortably, and a builder site stuffed with video and widgets can fail. Our WordPress performance guides test themes one by one, and DataCram's own custom builds on Next.js are sized for speed from the first page; they start at $1,500, fixed price, on our web design pricing page.

Google's own speed research, with the dates printed large

The most quoted speed numbers on the internet come from Google, and most are old. A September 2016 post from Google's DoubleClick team analyzed more than 10,000 mobile domains and found 53% of visits are likely to be abandoned when pages take longer than 3 seconds; the average mobile site then took 19 seconds to load over 3G. A 2017 Google and SOASTA study, updated in 2018, modeled a 123% rise in bounce probability as load time grows from 1 to 10 seconds, and a 95% drop in conversion probability as page elements grow from 400 to 6,000.

Two later Google figures measure the effect of its own policies. Its Search Central blog reported in April 2019 that after page speed became a mobile ranking factor in July 2018, the slowest third of traffic improved 15% to 20%, and abandonment of navigations from Search fell 20%. The Chrome team wrote in May 2020 that users were 24% less likely to abandon a page load, meaning leave before anything painted, on sites meeting the Core Web Vitals thresholds of the time.

On rankings, Google's page experience documentation, updated September 22, 2026, says Core Web Vitals are used by its ranking systems, that good results do not guarantee a top position, and that Google always seeks the most relevant content even when page experience is sub-par. Speed is a tiebreaker in a crowded local market, not a substitute for the right page; our technical SEO guide puts it in context.

Conversion studies: who measured, who paid, and how much each proves

Speed and sales studies range from controlled experiments to correlations sponsored by companies that want a faster web. Read the design column before the result column.

Portent's own data shows how fragile these curves are: its ecommerce sites at 1 second converted 2.5 times better than at 5 seconds but only 1.5 times better than at 10, which Portent puts down to a small sample of very slow pages. A curve that improves as the site gets slower is a curve with too few points. For the arithmetic on your own traffic, our essay on what a fast website is worth in dollars does it step by step, and landing page statistics covers conversion benchmarks.

The best speed study for your business is the one you run on your own checkout.

How to read these numbers before one lands in a pitch deck

1. 1

   Ask whether it is field or lab data

   The Chrome UX Report records real visits. A Lighthouse score is one simulated load on one device, useful for diagnosis and useless as a statistic.

2. 2

   Check who is in the sample

   Chrome's CrUX methodology counts only Chrome users who share usage statistics and sync history, excludes Chrome on iOS, and covers only sites that are public and popular enough. iPhone-heavy audiences are underrepresented.

3. 3

   Check the percentile

   Google grades the 75th percentile of page loads, split by mobile and desktop. An average load time hides the slow quarter of visits that the grade is built on.

4. 4

   Check the network year

   A figure measured on 3G in 2016 describes phones and pages that no longer exist. Use it for direction only.

5. 5

   Check the design and the sponsor

   An A/B test isolates speed; a correlation across brands does not. A study commissioned by a company that sells faster pages, or runs the ads they load, still counts, with the adjectives discounted.

Some famous figures are missing on purpose. The Amazon 100-millisecond anecdote and a widely repeated 2017 Akamai retail statistic could not be traced to a primary page we could open, so they stay out until we can see where they came from. They may be true. A statistic you cannot check is a rumor with a percent sign.

Your own speed numbers, and what to do with them

Every figure above is an average of other people's websites. Yours is free to look up: PageSpeed Insights shows your field data at the top when Chrome has enough visits, and Search Console's Core Web Vitals report groups slow pages by template. Our guides to how fast a website should load and why websites get slow explain the readings, and free SEO tools lists the rest of the kit.

For an outside read, the free website audit checks speed signals, SEO, mobile, security and content and emails a PDF with fixes, and the website health check runs 25 checks on one URL in about ten seconds. When the fix is bigger than a plugin, DataCram's site speed and Core Web Vitals service does the repair, and conversion rate optimization measures whether faster pages turned into more calls. Our SEO statistics page covers the search side of the same budget.