Website speed in 2026: more than half of mobile sites fail Google's test

Most business owners check their website on their own laptop, on office Wi-Fi, and decide it's fast. Your customers view it on a phone, on a tram, on LTE. Google measures the latter, and according to the latest data, more than half of mobile sites fail its test. Below are the numbers and what actually slows down WordPress sites.

Google measures your site's speed on your customers' phones, not on your laptop. Photo: Chris Liverani / Unsplash.

How many sites pass Core Web Vitals?

Core Web Vitals are three metrics Google uses to assess user experience: LCP (how fast the main content loads, good is up to 2.5 s), INP (how fast the page responds to a click, up to 200 ms) and CLS (whether elements jump around while loading). A site passes when all three are good for most real users.

According to the performance chapter of the Web Almanac 2025 (HTTP Archive, Chrome data from July 2025), 48% of sites pass on mobile and 56% on desktop. A year earlier it was 44% and 55%. It's better, but every other mobile site still fails. And the median time a phone is blocked running JavaScript grew from 1,209 to 1,916 ms, a 58% increase in one year.

Loading is the weakest point. The SEO chapter of the same report shows that only 62% of sites have good LCP on mobile, versus 74% on desktop. Responsiveness (INP) and layout stability (CLS) do much better.

WordPress and WooCommerce: where the problem is

I'm writing this as someone who builds on WordPress and WooCommerce every day, so no getting offended by the numbers. In the e-commerce chapter of the Web Almanac, WooCommerce stores pass Core Web Vitals on mobile 35% of the time. Shopify: 76%. Almost the whole difference is slow loading.

That isn't WordPress's fault as such, but of what gets piled on top of it. A PageSpeed Matters analysis (September 2026, top 1M sites) shows it very clearly:

• 0 to 2 detectable plugins: 57.9% of sites pass,
• 3 to 5 plugins: 45.9%,
• 6 to 8 plugins: 30.8%,
• 9 or more plugins: 24.0%.

Then there's page weight. According to the page weight chapter, the median mobile home page is 2.6 MB. Pages under 1 MB pass the test 57% of the time, those of 5 MB or more only 30%.

What does it cost in money?

The most cited study is Deloitte's "Milliseconds Make Millions" (37 brands, over 30 million mobile sessions). A speed improvement of just 0.1 seconds went hand in hand with 8.4% higher retail conversion and 9.2% higher average order value. A tenth of a second.

To be fair: it's a correlation on large brands, not a guarantee for every store. But I see the direction with clients every time. A faster site means fewer people leaving before they even see the offer.

What to check on your website

Before you pay for a new theme or a "speed-up plugin", go through these steps:

1. Check data from real devices. The "Core Web Vitals" report in Google Search Console or PageSpeed Insights (the real-user data section, not just the score). Look at mobile.
2. Count your plugins and third-party scripts. Every chat widget, pop-up, pixel, slider and page builder adds JavaScript. Remove anything you don't use or that doesn't make money.
3. Deal with the main image. It usually decides LCP. WebP or AVIF format, the right size, no lazy loading above the fold.
4. Check hosting and caching. A slow server response ruins everything after it. Good hosting, page caching and a CDN are usually the cheapest improvement.
5. Measure after every change. A new plugin or a campaign with another pixel can ruin a score you've built over months.

A fast website isn't a bonus, it's the baseline

You can have the best offer and the best ads, but if your site takes five seconds to load on a phone, some people will never see it. And Google sees that too.

If you want to know what exactly slows down your website, check out my technical SEO offer. I start with data from your Search Console and PageSpeed and finish with a prioritized list of fixes.

All the best,
Sławek

If your site exists but Google cannot see it, see the technical SEO offer .