How Website Speed Varies by Country: A Global Core Web Vitals Report

We measured how fast the web actually feels in 40 countries, using real-world performance data from millions of Chrome users rather than lab tests. For each country we calculated the share of websites that pass Google's Core Web Vitals, the three metrics Google uses to grade loading, responsiveness, and visual stability. We ran the numbers separately for mobile phones and desktops.

The headline result is a gap far wider than most teams expect. In Sweden, 72.4% of websites pass Core Web Vitals on mobile. In Nigeria, only 9.1% do. That is a spread of 63 percentage points across 40 countries that are all firmly online.

A few findings stood out across the full dataset:

• Sweden, Japan, and South Korea lead on mobile. Nigeria, Kenya, and South Africa sit at the bottom.
• Mobile is the harder surface almost everywhere. In 39 of the 40 countries, fewer sites pass on phones than on desktops. The average drop is 10.7 percentage points. Canada is the single exception, where mobile edges desktop by one point.
• Europe is the fastest region on mobile at 62.2%, while the Middle East and Africa group trails at 32.1%.
• The reason sites fail is consistent worldwide. Largest Contentful Paint, the loading metric, is the weakest of the three Core Web Vitals in all 40 countries.

The full country tables, the methodology, and the raw dataset follow. Everything here is free to cite and republish with attribution to PageSpeed Matters.

Most conversations about website speed start from a single assumption. A team runs a test from the office, sees a good score, and concludes the site is fast. The problem is that a website is not experienced in one place on one device.

It is experienced on a three-year-old Android phone on a crowded mobile network as often as on a wired laptop. The country a visitor sits in shapes that experience as much as the code behind the page.

Speed is not a vanity number. Google treats Core Web Vitals as a ranking signal, which means a slow site competes for search visibility with one hand behind its back (Google Search Central).

The commercial side is just as direct. Research from Portent found that a site loading in one second converts at 3.05%, while a site loading in five seconds converts at 0.41%. Deloitte measured that improving load time by 100 milliseconds lifted conversion rates by about one percent.

So the question "is the web fast" does not have one answer. It has roughly one answer per country, and the distance between the best and worst is the story this report sets out to measure.

We assessed more than 20.5 million country-level site records for the May 2026 CrUX release, about 13.0 million on mobile and 7.6 million on desktop. Here is where the 40 countries land on mobile, ranked by the share of sites that pass all three Core Web Vitals.

The shape of the table tracks wealth and infrastructure, though not perfectly. The top is held by richer economies with newer phones and faster networks. Sweden, Japan, South Korea, Germany, and the United Kingdom all cluster between 68 and 73 percent on mobile.

The bottom is held by markets where older devices and slower connections make loading the binding constraint. Nigeria, Kenya, South Africa, and Pakistan all sit under 25 percent.

The exceptions are the interesting part. The United States lands ninth on mobile at 63.3 percent, behind Sweden, Germany, the United Kingdom, Canada, the Netherlands, and Poland. A large, device-diverse market carries more slow experiences than a smaller and more uniform one.

At the other end, Nigeria reports that the median site loads its main content in 4.4 seconds on a phone, against 1.5 seconds in Sweden. The same web, built with the same frameworks and the same content management systems, arrives nearly three seconds later for a Nigerian visitor.

Split every country by device and a clear pattern appears. In 39 of the 40 countries, the Core Web Vitals pass rate is lower on mobile than on desktop. The average gap is 10.7 percentage points. The only country that breaks the rule is Canada, where mobile actually comes in one point ahead of desktop.

The widest gaps land in fast-growing, mobile-first markets, where office desktops skew toward newer machines while phones skew old and cheap.

The narrowest gaps belong to mature markets. The United States loses 1.6 points moving from desktop to mobile, Sweden 0.8, and the United Kingdom 0.7. In those countries the typical phone and the typical network are close enough to a desktop that the page barely notices the change of device.

For a business, this is the most practical line in the entire study. The desktop score a team is proud of is not the score most of its customers feel. A Brazilian retailer looking at a 55.6% desktop pass rate is serving a 34.7% reality to the phone-first majority of its market.

Grouping the 40 countries into regions shows where the fast web is concentrated and where it thins out. These are average mobile pass rates across the countries we measured in each region.

Europe's lead rests on depth rather than a single standout. A dozen countries, almost all above 53 percent on mobile, pull the regional average up. Sweden, Germany, the United Kingdom, the Netherlands, and Poland anchor the top, and even the slower European entries such as Italy and Russia sit well above the global midpoint.

Asia is the most divided region in the study. Japan, South Korea, and Taiwan rank among the fastest places on earth for mobile web, while Bangladesh, the Philippines, and Pakistan rank among the slowest. The regional average of 44.2% hides that split entirely.

The Middle East and Africa group sits last, weighed down by Nigeria, Kenya, and South Africa, where loading times on mobile stretch past three and a half seconds for the median site. South America is only slightly ahead, with Brazil, Argentina, Colombia, and Chile all between 33 and 41 percent on mobile.

In both regions, the pattern is the same: desktop performance is respectable, and mobile performance falls off a cliff.

A few countries show, in miniature, the forces shaping the whole table.

Japan finishes second on mobile at 70.1% and first on desktop at 77.0%, but the more telling number is its visual stability. Japanese sites post the highest good-CLS rate in the study at 92.2% on mobile, and the median Japanese site records a desktop interaction time of just 25 milliseconds.

The Japanese web is not only fast on average. It is consistent, with fewer of the layout jumps and sluggish taps that drag other markets down.

The United States is the clearest example of scale working against a country. It ranks ninth on mobile at 63.3%, yet its mobile penalty is one of the smallest anywhere at 1.6 points. The country is not losing ground on phones. It is simply averaging across an enormous and varied population of devices.

Brazil is the case study in the mobile penalty. Its desktop pass rate of 55.6% is respectable. Its mobile pass rate of 34.7% tells a different story, a fall of 20.9 points and one of the steepest drops in the study.

For the largest digital economy in South America, the gap between how the web performs in the office and how it performs in the hand is the single biggest performance issue, and it sits almost entirely on the loading metric, where only 50.5% of Brazilian sites are good on mobile.

A site fails Core Web Vitals if it fails any one of the three metrics. Looking across all 40 countries, the weakest metric is the same everywhere. Largest Contentful Paint, the measure of how quickly the main content of a page loads, is the lowest-scoring of the three Core Web Vitals in every single country, on mobile and desktop alike.

The numbers make the case plainly. In Nigeria, only 13.6% of sites have good LCP on mobile, while 40.5% have good responsiveness and 81.4% have good visual stability.

Even in fast markets the order holds. Sweden posts 84.8% good LCP against 90.9% good responsiveness and 88.0% good visual stability. Wherever you look, loading is the metric dragging the pass rate down.

Visual stability, measured by Cumulative Layout Shift, is close to a solved problem. The median site reports a p75 layout shift of effectively zero in every country we measured, and good-CLS rates run from 73% in the Philippines up to 92% in Japan.

Responsiveness, measured by Interaction to Next Paint, is where the hardware gap shows itself most sharply. INP replaced First Input Delay as a Core Web Vital in March 2024, and it is far more sensitive to a slow processor because it captures how long the page takes to respond after a tap.

In wealthy markets, responsiveness is excellent: Sweden, Germany, the United Kingdom, and Japan all clear 89% good INP on mobile. In markets running on cheaper phones it falls apart, dropping to 40.5% in Nigeria, 50.1% in Kenya, 52.4% in the Philippines, and 56.6% in Bangladesh.

On desktop, responsiveness almost disappears as a problem. Good-INP rates run between 95 and 99% across the table, and the median desktop interaction completes in 25 to 100 milliseconds.

The instruction that falls out of this is short. If a team can only fix one thing, it should fix loading. Heavy images, render-blocking scripts, and slow server responses are what keep the global pass rate down.

Three neutral takeaways follow from the country tables.

First, the mobile score is the one that matters. In 39 of 40 countries, mobile trails desktop, often by double digits. A dashboard showing a green desktop badge is not evidence that most of a site's visitors are having a fast experience.

Second, loading is the metric to fix first. LCP is the weakest score in every country on both devices. Heavy hero images, render-blocking scripts, and slow server responses are the recurring culprits, and they respond well to focused image, caching, and delivery work.

Third, geography matters. A store serving mid-range devices in Indonesia, the Philippines, or Nigeria needs to be materially lighter than a competitor serving Sweden or Germany just to reach the same field score. That is a page-weight problem, not a metric problem, and the fix lives in the build.

A study earns citations by being reproducible, so here is exactly what we did.

Data source. We used the Chrome User Experience Report, Google's public dataset of real-world performance collected from opted-in Chrome users. We queried the chrome-ux-report.materialized.country_summary table on BigQuery for the May 2026 release. This is field data, meaning it reflects what real people on real devices and networks actually experienced.

What we measured. We assessed the three Core Web Vitals at the 75th percentile, using Google's official thresholds for a "good" experience: LCP at or under 2,500 milliseconds, INP at or under 200 milliseconds, and CLS at or under 0.10. A website passes Core Web Vitals when its 75th-percentile experience is good on all three metrics at once.

How we built each country's score. For every country we calculated the share of websites that pass, weighting each website equally. This mirrors the approach used in the HTTP Archive Core Web Vitals Technology Report. We also recorded each country's median site, defined as the median across all sites of their individual 75th-percentile values.

Scope. The study covers 40 countries across every populated region, split by device into mobile (phone) and desktop. The sample comprises 20,572,969 country-level site records, of which 12,976,793 are mobile and 7,596,176 are desktop.

Caveats, stated plainly. CrUX reflects Chrome users only. In markets where Chrome is not the dominant browser, most notably China, the dataset is not representative of the whole web, so we excluded those markets. A country's score reflects the sites its users visit, which includes global sites, so it measures the lived experience of that country's users rather than only the sites built locally. Very small sites and very small countries are excluded by CrUX itself for privacy and sample-size reasons.

One note on the median CLS column. Median CLS reads as 0.00 in almost every country. That is not a sign that every site passes. It reflects a heavily skewed distribution in which more than half of all sites have near-zero layout shift, so the middle site sits at zero while a failing tail still exists. The "good CLS" share, which ranges from 73 to 92 percent, is the figure that captures the pass rate.

Reproduce it. The full SQL is published alongside this report. Anyone with a free Google Cloud account can re-run it. The headline query scans roughly 120 gigabytes, which sits inside BigQuery's free monthly tier.

This edition is the baseline. Because the study is built entirely on the public CrUX dataset, every figure can be recomputed each month, so future editions will be able to track which countries are speeding up and which are slipping.

The places worth watching are at the bottom of today's table. As mid-range Android hardware improves and content delivery networks push closer to users, the countries with the most room to move, Nigeria, Kenya, and South Africa, are where the largest swings are most likely to appear.

This dataset is free to cite and republish with attribution to PageSpeed Matters and a link to this report. The raw CSV, the full mobile and desktop tables, and an interactive country table are available in the press kit.

Suggested citation: PageSpeed Matters, How Website Speed Varies by Country: A Global Core Web Vitals Report, June 2026. Data: Google Chrome User Experience Report, May 2026 release.

• Core Web Vitals data: Google Chrome User Experience Report on BigQuery, chrome-ux-report.materialized.country_summary, May 2026 release. developer.chrome.com/docs/crux
• Core Web Vitals thresholds and ranking-signal status: Google Search Central
• Conversion rate by load time: Portent, Site Speed Is Hurting Everyone's Revenue.
• Conversion lift per 100ms of load time: Deloitte, Milliseconds Make Millions.

Appendix: desktop ranking (top and bottom 10)