Load the same website on a phone in the US and on a phone in Nigeria. Chrome's own data says the US visit will pass Google's speed bar 78 percent of the time. The Nigerian one: 13 percent. Same website, same code, same server. Different visitor.
We looked at six years of Chrome's real-user data, every September from 2020 to 2026, across 12 countries. We wanted to know three things: why the rest of the world loads slower, whether six years of a faster web reached everyone, and what a site owner can do about it. This is original research from the Chrome UX Report (CrUX), Google's measurement of how real Chrome users experience websites. Method and controls are at the end.
Two honesty notes. This is Chrome on phones only, so iPhones and other browsers are not in it. And the pass rates are shares of websites, not shares of visits: a site counts as passing when its typical visit is fast enough.
What 'pass' means here
Google measures how long the main content of a page takes to show up on a real visitor's screen. That measure is called Largest Contentful Paint. A site passes when its typical visit (the 75th percentile, Google's own cut) shows the main content within 2.5 seconds. We counted the share of websites that pass, per visitor country.
Same website, different country
To keep this fair, we held the websites still. We took the sites Chrome measured for both US visitors and visitors in each other country, in every September from 2020 to 2026, and compared the two groups of visitors on exactly the same sites.
The website did not change between the two dots. The visitor did, and with them their network, their phone, and how far they sit from the server.
Table view
| Visitors in that country | US visitors to the same sites | |
|---|---|---|
| Nigeria | 13.3% | 77.9% |
| Pakistan | 29.3% | 76.8% |
| South Africa | 35.6% | 79.6% |
| India | 37.1% | 78.0% |
| Indonesia | 50.3% | 77.8% |
| Australia | 56.7% | 78.2% |
| Brazil | 57.8% | 78.9% |
| Japan | 64.6% | 73.1% |
| Mexico | 64.7% | 76.8% |
| Germany | 78.8% | 73.0% |
| United Kingdom | 81.0% | 76.8% |
Sites with data for both US visitors and that country's visitors in every September 2020 to 2026 (5,997 sites for Nigeria up to 40,676 for the UK). Source: Chrome UX Report country data, phone visits; analysis by Stackra.
About half the gap arrives before the page starts
The first thing a page waits for is the server's first reply, called time to first byte. That wait is mostly distance and round trips across the network. On the same sites, roughly half of the extra load time for visitors outside the US lands before that first byte. The rest comes after it: slower downloads and slower phones drawing the page.
| Country | Extra load time | Of which before the first byte | Share |
|---|---|---|---|
| Nigeria | 2,200 ms | 1,000 ms | 45% |
| Pakistan | 1,300 ms | 600 ms | 46% |
| South Africa | 1,200 ms | 500 ms | 42% |
| India | 1,100 ms | 600 ms | 55% |
| Indonesia | 800 ms | 300 ms | 38% |
| Brazil | 600 ms | 300 ms | 50% |
| Australia | 600 ms | 400 ms | 67% |
| Mexico | 400 ms | 100 ms | 25% |
| Japan | 300 ms | 200 ms | 67% |
Median per-site difference, local visitors minus US visitors, on the same websites. Bandwidth and phone speed are not in Chrome's data, so the part after the first byte is attributed to them by elimination.
Six years faster, mostly in the page
On the same websites, the share passing for US visitors went from 46 percent in 2020 to 81 percent in 2026. India rose from 16 to 44 percent. Nigeria rose from 5 to 14 percent and has stalled since 2024.
Table view
| US | India | Nigeria | |
|---|---|---|---|
| 2020 | 45.6% | 15.7% | 4.9% |
| 2021 | 56.1% | 26.0% | 6.3% |
| 2022 | 61.7% | 27.7% | 8.6% |
| 2023 | 67.7% | 29.0% | 9.7% |
| 2024 | 77.2% | 39.2% | 15.9% |
| 2025 | 78.5% | 35.7% | 15.2% |
| 2026 | 81.4% | 43.6% | 13.9% |
Websites with data in every September, per country: US 359,724, India 79,631, Nigeria 7,979. Source: Chrome UX Report country data, phone visits; analysis by Stackra.
Where the time went
Almost all of that gain came after the server's first reply. On the same US sites, the server answered a little sooner, 800 to 700 milliseconds. Everything after it, the part where the page downloads and draws, dropped from 1,700 to 800 milliseconds. India shows the same shape. Lighter pages, faster browsers and faster phones all show up in that second part, and this data cannot tell them apart.
Table view
| US: Waiting for the server's first byte (ms) | US: Loading after the first byte (ms) | India: Waiting for the server's first byte (ms) | India: Loading after the first byte (ms) | |
|---|---|---|---|---|
| 2020 | 800 | 1,700 | 1,600 | 2,300 |
| 2021 | 800 | 1,400 | 1,400 | 1,800 |
| 2022 | 800 | 1,300 | 1,500 | 1,700 |
| 2023 | 800 | 1,100 | 1,500 | 1,600 |
| 2024 | 700 | 900 | 1,400 | 1,400 |
| 2025 | 700 | 900 | 1,500 | 1,400 |
| 2026 | 700 | 800 | 1,300 | 1,300 |
Each part is a median across the same websites, so the two parts do not add up to an exact load time. Source: Chrome UX Report country data, phone visits; analysis by Stackra.
The network barely moved
If faster networks explained the gains, the share of visits on 4G should have climbed with them. In the US it did not move: about 96 percent of visits on these sites ran on 4G in both 2020 and 2024, while the US pass rate rose more than 30 points. India's 4G share rose, mostly in 2021, which explains part of its first jump. Nigeria's 4G share peaked in 2022 and then fell back.
Table view
| US | India | Nigeria | |
|---|---|---|---|
| 2020 | 96.4% | 74.6% | 61.2% |
| 2021 | 96.2% | 83.8% | 70.8% |
| 2022 | 95.8% | 82.4% | 71.1% |
| 2023 | 96.5% | 84.6% | 66.7% |
| 2024 | 96.9% | 86.1% | 63.2% |
The axis starts at 50 percent. Chrome changed how it labels connection types in January 2025, so the series stops at 2024; since then, round-trip times on these sites barely moved (Nigeria 325 ms in both 2025 and 2026). 4G is Chrome's top label, so 5G counts as 4G.
One jump belongs to Chrome, not to websites
Between 2023 and 2024, every country jumped at once, by 4 points in Japan up to 16 in Brazil, for US and local visitors alike, while server response times barely moved. A move that uniform points to the browser. Chrome's own release notes for March 2024 credit a change to cookie caching in Chrome for Android with noticeable gains in loading speed. Part of that year's improvement was the browser getting faster, not websites.
The middle is catching up. The bottom is not.
Here is the gap between US visitors and local visitors on the same websites, in 2020 and in 2026. Where networks are decent, lighter pages helped everyone and the gap closed. Where networks are slow, lighter pages mostly helped the visitors who were already close to passing.
In Nigeria, the server's first reply arrives about a second later than it does for US visitors to the same sites, and that delay grew from 800 to 1,000 milliseconds over six years. The pass rate for Nigerian visitors on these sites topped out at 16 percent in 2024 and was 13 percent in 2026.
Table view
| Gap in 2020 | Gap in 2026 | Change | |
|---|---|---|---|
| Nigeria | 39.3 | 64.6 | +65, wider |
| Pakistan | 32.6 | 47.5 | +48, wider |
| South Africa | 35.3 | 44.0 | +44, wider |
| India | 32.6 | 40.9 | +41, wider |
| Indonesia | 27.5 | 27.5 | +28, flat |
| Australia | 17.6 | 21.5 | +22, wider |
| Brazil | 29.1 | 21.1 | +21, narrower |
| Mexico | 17.6 | 12.1 | +12, narrower |
| Japan | 2.0 | 8.5 | +9, wider |
| United Kingdom | -7.6 | -4.2 | -4, narrower |
| Germany | -9.5 | -5.8 | -6, narrower |
Below zero, local visitors pass more often than US visitors. Same websites in every year, so a change here is not caused by sites joining or leaving the data. Source: Chrome UX Report country data, phone visits; analysis by Stackra.
Where the server sits
Inside a single country, sites on that country's own domain (like .com.au or .de, a rough sign they are hosted nearby) pass more often than everything else. In Nigeria it makes no difference: when the network itself is the limit, a closer server cannot fix it alone.
Table view
| Sites on the country's own domain | All other sites | |
|---|---|---|
| Nigeria | 15.5% | 14.8% |
| South Africa | 39.3% | 33.8% |
| Indonesia | 52.8% | 52.7% |
| India | 53.4% | 42.2% |
| Brazil | 56.7% | 44.6% |
| Mexico | 60.1% | 56.4% |
| Australia | 71.4% | 51.5% |
| Germany | 85.4% | 74.2% |
| Japan | 88.8% | 79.4% |
A country domain is a rough stand-in for hosting nearby, not proof of where the server is. Source: Chrome UX Report country data, phone visits; analysis by Stackra.
What to do with this
The practical read depends on where your visitors are.
- •Look at where your visitors are, not at a global average. Chrome reports speed per country. If a meaningful share of your visitors are in India, South Africa or Pakistan, your US number is not their experience.
- •For far-away or slow-network audiences, start with distance: a server or CDN near them, and fewer round trips before the first byte. Then trim the page.
- •Know the floor. On the slowest networks, no single change gets a typical visit under 2.5 seconds. That is worth knowing before you chase a number you cannot reach.
Baselines and controls
So you can weight the read, here is what this comparison holds constant and what it does not.
| Held constant | Not controlled |
|---|---|
| The same websites in every year for each trend (Chrome roughly doubled its coverage in 2022) | Bandwidth and phone speed: Chrome does not report them |
| The same websites for US and local visitors in the gap comparison | iPhones and non-Chrome browsers: not in the data |
| One pass rule everywhere: typical visit within 2.5 seconds | Where each server actually sits: the country domain is a proxy |
| The same month every year (September) | The 2023 to 2024 browser change is credited to Chrome by its own release notes, not measured separately |
Connection type is kept to 2020 to 2024 because Chrome changed how it labels connections in January 2025.
Methodology
Data comes from the Chrome UX Report's country table (country_summary), phone visits, every September from 2020 to 2026, for 12 countries. A site passes when its 75th percentile Largest Contentful Paint is 2,500 milliseconds or less. Trends use only websites with data in all seven Septembers for that country. The US vs local comparison uses only websites with data for both US visitors and that country's visitors in all seven years. Time to first byte and the time after it are medians of each site's 75th percentile values. The 4G share is the share of visits Chrome labels 4G out of all labelled connections.
Frequently asked questions
Common questions about why websites load slower outside the US.
Why do websites load slower outside the US?
Mostly because of the visitor's network, not the website. On the same websites in September 2026, 78 percent passed Google's loading threshold for US visitors but 37 percent for visitors in India and 13 percent in Nigeria. About half of the extra time arrives before the server's first reply, which is distance and network round trips; the rest comes after it.
Has the web gotten faster since 2020?
Yes, on the same websites, everywhere we measured. The share passing for US visitors rose from 46 to 81 percent between 2020 and 2026, India from 16 to 44 percent, and Nigeria from 5 to 14 percent. Most of the gain came after the server's first reply, in how fast pages download and draw.
Is the speed gap between countries closing?
In the middle, yes; at the bottom, no. On the same websites, the gap between US and local visitors narrowed from 2020 to 2026 in Brazil (29 to 21 points) and Mexico (18 to 12), stayed flat in Indonesia (28), and widened in India (33 to 41), South Africa (35 to 44), Pakistan (33 to 48) and Nigeria (39 to 65).
Does hosting a website closer to visitors make it faster?
Where networks are decent, it helps. Sites on a country's own domain, a rough sign of local hosting, passed more often than other sites in Australia (71 vs 52 percent), Germany (85 vs 74), India (53 vs 42) and Japan (89 vs 79). In Nigeria it made no difference (16 vs 15 percent), because the network is the limit.
Did faster mobile networks cause the speed gains?
Only partly. In the US, about 96 percent of visits ran on 4G in both 2020 and 2024 while the pass rate rose more than 30 points, so most of the gain came from websites and browsers. India's 4G share rose from 75 to 86 percent, which explains part of its gains. Nigeria's 4G share fell after 2022.
Why did every country get faster in 2024?
Largely because Chrome did. Every country jumped between 2023 and 2024, by 4 to 16 points, for US and local visitors alike, while server response times barely moved. Chrome's release notes for March 2024 credit a cookie caching change in Chrome for Android with noticeable loading gains.