TL;DR

Website performance benchmarks in Mainland China look very different from global averages. While Google recommends an LCP under 2.5 seconds and a TTFB under 0.8 seconds, most global websites don't come close to meeting those thresholds when tested from China. Chinafy's research, based on testing 614 global websites from Beijing, Virginia, and London, found that 2 in 3 global websites failed to load in China, and TTFB was 4 to 4.5x higher in Beijing than in the US or UK. These numbers reflect the reality of cross-border internet infrastructure and third-party resource incompatibilities, not poor website development. The good news is that with the right optimization, global websites can achieve near-native onshore performance without a full rebuild. Read the full report here.

What does "good" website performance look like globally?

Before looking at China-specific benchmarks, it helps to establish a baseline. Google's Core Web Vitals, which are used as ranking signals and as general indicators of user experience, set the following thresholds for "good" performance:

Largest Contentful Paint (LCP): under 2.5 seconds

Interaction to Next Paint (INP): under 200 milliseconds

Cumulative Layout Shift (CLS): under 0.1

Time to First Byte (TTFB): under 0.8 seconds (not a Core Web Vital, but a closely related performance metric)

These are the benchmarks most web teams use to evaluate their site's speed and responsiveness. According to Google's own data, only about 47% of websites currently meet all Core Web Vitals thresholds globally. That means more than half of all websites already fall short of "good" performance, even before factoring in the added complexity of delivering content into China.

47% of users also expect a website to load in two seconds or less, and research from Google shows that 53% of mobile users abandon sites that take over 3 seconds to load.

Why global benchmarks don't tell the full story in China

There's an important nuance when applying global performance standards to China.

China's domestic internet is fast. Major Chinese platforms like Baidu load in well under a second from within the country. Chinese internet users, of which there are over 1.1 billion, are accustomed to this speed. Research suggests that 73% of mobile users in China expect a website to load in under three seconds, a higher expectation than the 58% recorded in the US.

But global websites loading into China face a very different set of conditions. Cross-border traffic passes through a limited number of international gateways, which adds latency and increases the chance of packet loss. On top of that, many of the third-party services that global websites rely on (Google Fonts, YouTube, Facebook plugins, and hundreds more) are slow or inaccessible from within China.

So while China's domestic web may be among the fastest in the world, the experience of accessing a global website from China is often the opposite.

How global websites actually perform in China

Chinafy's State of Global Website Performance in China report, based on testing 614 global websites from Beijing (China), Virginia (US), and London (UK), puts concrete numbers on the gap.

The headline findings:

2 in 3 global websites failed to load in China

TTFB was 4 to 4.5x higher in Beijing than in Virginia or London

Most websites loaded 10 to 30x slower in China than in other regions

Many failed to render key content at all

To put this in context, a website that loads in 2 seconds from the US or UK might take 20 seconds to load from Beijing, if it loads at all. At those speeds, most visitors will have left the page long before it finishes rendering.

Some industries showed particularly high failure rates. Hotel websites, for example, had the highest failure rate of any industry tested, with 75% timing out completely when tested from Beijing. The ones that did load averaged 22.9 seconds.

These reflect a pattern across industries where global websites depend heavily on third-party resources that don't work reliably in China.

Performance benchmarks by metric

Here's how the key metrics compare when testing from China versus testing globally.

Time to First Byte (TTFB)

Globally, a good TTFB is under 0.8 seconds. From China, TTFB for global websites is typically 4 to 4.5x higher than the same site tested from the US or UK. For an unoptimized global site, TTFB from Beijing can range from 3 to 8+ seconds, depending on hosting location and network conditions.

Largest Contentful Paint (LCP)

Google's "good" threshold is 2.5 seconds. Most global websites tested from China exceed this by a wide margin. When third-party render-blocking resources (fonts, scripts, analytics libraries) fail to load or take extended periods to time out, LCP can stretch well beyond 10 seconds.

Page Complete

Page Complete measures when a page has finished loading all resources. Globally, this might range from 3 to 8 seconds for a typical site. From China, global websites that do eventually load often take 15 to 30+ seconds to reach Page Complete, with many never completing at all.

Resource delivery rate

A metric that's less commonly discussed but highly relevant in China is the percentage of a page's resources that actually get delivered. For global websites tested from China, it's common to see only 60 to 70% of resources load successfully. The missing 30 to 40% are typically third-party resources that are blocked, timed out, or severely delayed.

The takeaway is that standard global benchmarks (LCP under 2.5s, TTFB under 0.8s) are still good targets. Websites serving visitors in China should aim for these same standards, not settle for far slower performance just because the environment is more challenging.

What about domestic Chinese websites?

It's worth noting the contrast. Domestic Chinese websites, built with China-native infrastructure and without reliance on foreign third-party services, perform well within the benchmarks that Chinese users expect.

Baidu, for instance, loads in approximately 0.6 seconds from within China. Other major platforms like Taobao, JD.com, and WeChat's web services perform similarly.

This is the speed that Chinese internet users experience daily. When they then visit a global website that takes 20 seconds to load, or shows a partially rendered page with missing fonts and broken forms, the gap in experience is significant. They're not comparing your site to other global sites loading into China. They're comparing it to the local platforms they use every day.

This is why "it's slow for everyone in China" isn't a strong position. Chinese users already know what fast looks like.

Lower benchmarks don't mean lower standards

There's a temptation, when faced with the reality of China's cross-border internet conditions, to accept that slower performance is just how things are for global websites. Benchmarks specific to "global sites in China" might suggest that 10 seconds is a reasonable load time because it's better than the 17 second average.

This thinking is understandable, but it can work against you.

A visitor in China who encounters a 10-second load time isn't thinking "well, that's pretty good for a foreign website." They're thinking about whether to wait or leave. The bounce rate data makes this clear: pages that take 5 seconds to load see bounce rates as high as 38%, and beyond that, the drop-off accelerates. Research from Google shows that a one-second delay in mobile load times can reduce conversion rates by up to 20%.

The performance environment in China may be more difficult, but the expectations of users in China are not lower. If anything, they're higher, given how fast domestic platforms perform.

The practical goal for a global website in China is to get as close to the same experience as possible. That means aiming for near-native onshore performance, not just "better than most foreign sites."

Why third-party resources are the biggest factor

When companies first learn about performance issues in China, the initial assumption is usually that it's a hosting or CDN problem. Move the site closer to China, add a CDN with regional coverage, and the issue goes away.

Hosting location and CDN do matter for first-party content delivery. But for most global websites, the third-party resources are what account for the majority of performance degradation in China.

A typical website loads resources from an inexhaustible list of external domains. Google services alone (Fonts, Analytics, Tag Manager, Maps, reCAPTCHA) might account for 5 to 15 separate resource requests. Add in social media plugins, embedded video players, live chat widgets, A/B testing tools, and marketing automation scripts, and the number of external dependencies grows quickly.

In China, many of these requests either fail entirely or take so long to respond that they stall the page. A render-blocking Google Maps request might hold up the page for 30 seconds before the browser gives up and moves on.

This is why a CDN alone, which only speeds up delivery of your own content, doesn't fully address the performance gap. The code-level incompatibilities at the application layer need to be handled separately.

How to benchmark your own site in China

If you're trying to understand how your website performs in China, there are a few approaches:

Chinafy's free speed test tool lets you test your site from China and compare it against global load times

WebPageTest supports testing from Beijing, giving you a waterfall view of how each resource loads (or doesn't)

Get in touch with Chinafy for a China Web Performance Audit with actionable recommendations for how to fix it.

When testing, pay attention to more than just the headline load time. Look at which specific resources are failing or timing out, what percentage of your page's assets are being delivered, and where the biggest delays occur in the loading waterfall. These details will tell you far more about what's actually going wrong than a single speed score.

How Chinafy helps

Chinafy optimizes global websites for China at both the infrastructure and code level. Rather than requiring a rebuild, rehost, or migration, Chinafy works as a bolt-on solution that detects and handles the third-party resources causing performance issues, and layers on infrastructure built for the region, including near-China or onshore CDN delivery.

Websites optimized with Chinafy typically see significant improvement in performance metrics like Page Complete, DOMComplete, and TTFB, and perform better compared to using a CDN alone. Results vary depending on the site's composition, regional network conditions, and other external factors.

For a detailed look at how global websites perform in China across industries, testing methodology, and what good performance looks like in the region, read the full report:

Read Chinafy's State of Global Website Performance in China report

FAQs

What is a good website load time in China? 

The same benchmarks that apply globally still apply in China: an LCP under 2.5 seconds, TTFB under 0.8 seconds, and Page Complete under 5 seconds. While most global websites currently fall far short of these when tested from China, optimized sites can achieve near-native onshore performance and load fully in 3 to 5 seconds.

Why do global websites load so slowly in China? 

The two main factors are cross-border latency (traffic passing through limited international gateways) and third-party resource incompatibilities (services like Google Fonts, Google Analytics, YouTube, and Facebook being slow or inaccessible in China). A CDN addresses the first factor but not the second.

How do I test my website's performance in China? 

You can use Chinafy's free speed test, WebPageTest (which supports testing from Beijing), or services like 17ce.com for multi-province testing. Look at individual resource loading in the waterfall, not just the headline number.

Is it normal for websites to fail to load in China? 

Unfortunately, yes. Chinafy's research found that 2 in 3 global websites failed to load when tested from Beijing. This is common for sites that haven't been specifically optimized for the region, and it's driven primarily by third-party resource incompatibilities.

Should I accept slower performance in China because the environment is harder? 

No. While the environment does present additional challenges, users in China are accustomed to fast domestic platforms and have high performance expectations. A global site loading in 10 to 15 seconds when domestic alternatives load in under a second will lose visitors. The goal should be near-native onshore performance, not just incremental improvement.

Do I need to host in China to get good performance there? 

Not necessarily. Onshore hosting reduces latency but doesn't address third-party resource incompatibilities, which are often the larger contributor to slow load times. An offshore website optimized at both the code and infrastructure level can achieve near-native onshore performance without an ICP license or onshore hosting migration.