What makes a website fast

Google measures three things on every page it has enough data for, using real visits from Chrome users rather than a test in a lab. Together they are called Core Web Vitals.

Largest Contentful Paint is how long it takes for the biggest thing on the screen, usually a heading or an image, to appear. Google calls 2.5 seconds or under good, and over 4 seconds poor.

Interaction to Next Paint is how long the page takes to respond when a visitor taps or clicks. 200 milliseconds or under is good; over 500 is poor. It replaced an older measure on 12 March 2024.

Cumulative Layout Shift is how much the page jumps about while it loads, the thing that makes you tap the wrong link. A score of 0.1 or under is good; over 0.25 is poor.

A page passes when it meets all three for the slowest quarter of its visitors, not the average visitor on a fast connection. Mobile phones and desktops are scored separately, and mobile phones are where business sites fail.

When we say a site we build is fast, we mean that on a mobile phone, for three visitors in four, the main content is on screen in under two seconds. That is stricter than Google’s 2.5 seconds, and you can check it.

Where Google has enough real visitors to report field data, that is the measure. Where it does not, we run the lab test in Google’s PageSpeed Insights several times on its standard mobile setting and quote the median, and we say which one a figure comes from.

Our own measurements on 28 October 2026 put the homepage of OpenThrottle at 1.2 seconds, IFS Coatings at 1.4 seconds and Matthew Brook at 1.1 seconds on that mobile test. The method is beside each figure in the case studies.

In July 2025, 48 per cent of the websites in the HTTP Archive’s sample passed all three Core Web Vitals on mobile. The same year’s report puts sites running WordPress at 45 per cent. So a business owner who assumes their site is probably fine has slightly worse than even odds.

The median mobile page in that year’s report weighed 2,164 kilobytes and made 72 requests. A page that size is not going to be usable in two seconds on a mid-range mobile phone. Weight is not the whole story. The IFS Coatings homepage carries large product images and still paints its main content in 1.4 seconds, because the visitor is not made to wait for them. But weight is where the trouble usually starts.

Google says Core Web Vitals are used by its ranking systems, that there is no single page-experience signal, and that a relevant page still wins over a faster one. In its own words, “trying to get a perfect score just for SEO reasons may not be the best use of your time.”

We take that at face value. The stronger reasons to be fast are the visitor who leaves, the form that never gets filled in, and the fact that a slow site is usually a heavy one carrying things that also make it insecure.

Most agencies quote one figure: 53 per cent of visits to a mobile page are abandoned if it takes more than three seconds to load. It comes from Google data on about 3,700 mobile sites in March 2016, and we have found nothing newer that replaces it, so nobody should quote it as current. A 2020 study by Deloitte Digital and 55, run with Google, of 37 brands and 30 million sessions, linked a tenth of a second’s improvement to 8.4 per cent more conversions on retail sites. That is a correlation across sites, not a test, and for lead-generation sites form completions rose 21.6 per cent while the measured conversion rates fell.

We have found only two controlled experiments that tie Core Web Vitals to business results, and both are on consumer sites. Vodafone in 2021, on one landing page: a 31 per cent improvement in time to main content produced 8 per cent more sales and 15 per cent more leads per visit. Rakuten in 2022: a 33.13 per cent increase in conversion rate. We have found no study of speed on industrial or manufacturing sites, so any figure a provider gives you for your sector has been carried over from retail. The evidence is thinner than the industry says. The case for speed stands on Google’s thresholds, the visitor who will not wait, and what a slow page is carrying.

PageSpeed Insights also shows a performance score from 0 to 100, produced by a Google tool called Lighthouse after loading the page once on a simulated slow mobile phone. It is useful for finding what to fix. It is not the same as passing Core Web Vitals, because it is one lab load, it cannot measure how the page responds to a tap, and it varies from run to run.

So we treat the score as a check before launch and the field data as the result. A site can score 100 and still fail on real mobile phones. We would rather tell you that than quote the score.

1. Go to pagespeed.web.dev and enter your homepage address.
2. Make sure the Mobile tab is selected.
3. Read the top section first. If it says “Core Web Vitals Assessment: Passed” or “Failed”, that is Google’s real visitor data for your site over the past 28 days. If it says there is not enough data, your site is too small for field data and the lab figures below are all you have.
4. Note the Largest Contentful Paint figure. Under 2.5 seconds is Google’s pass. Under 2 seconds is ours.
5. Do the same for the pages your customers actually use, such as the product list and the contact page.

If you have access to Google Search Console, its Core Web Vitals report does the same for every page it has data on and shows you which pages are dragging the site down.

The same handful of things, in the same order of frequency.

Images that are uploaded at full camera size and shrunk on the page, so the mobile phone downloads ten times what it shows. Scripts added over the years and never removed: a chat widget, a tag manager, a slider, two analytics tools, a cookie banner. Each is small; together they stop the page painting. Fonts loaded from a third party before any text can appear. On WordPress sites, a theme and a dozen plugins each adding their own stylesheet and script to every page whether the page uses them or not, and a server that builds each page from a database on every visit.

None of these is fixed downstream. A faster host or a caching plugin can only pass on what the page gives it. The fix is a page that carries only what the visitor needs, in the order they need it.

Every page on this site that says fast means this. The Websites page builds to it, the Website Risk Analysis measures your site against it, and the case studies quote it with the method beside the figure.

If your site failed the five-minute check, or you could not tell, talk to an engineer. Tell us the address and what the site needs to do. We will tell you plainly what it would take.