Why every second of load time counts
A second feels like nothing. But someone staring at a blank screen on a phone with a slow connection may well go straight back to the search results. This article explains how fast is fast enough, and how to measure it.
Google calls a page good when, for at least three out of four real visits, the largest content is on screen within 2.5 seconds, the page responds to a tap within 0.2 seconds and it hardly shifts. In our own measurement on a phone, the homepages of our sites in Astro and Nuxt were on screen after 0.58 to 1.63 seconds.
When someone opens a page on a phone, on the train or out in the countryside, they first see nothing. Your visitor notices how long that takes before you do, because you usually open your own site on a fast laptop.
In this article you will find out what Google means by fast, what slowness costs, how fast the sites we built are, where the delay often comes from and how to measure your own site yourself.
How fast should a website load?
Google measures the experience of a page with three metrics, the Core Web Vitals. For each metric Google sets a threshold for good and a threshold for poor. A page counts as good when, for at least three out of four visits (the 75th percentile), all three metrics stay within the threshold, measured on real visitors and separately for phone and desktop.
| Metric | What it measures | Good | Poor |
|---|---|---|---|
| LCP | How long it takes until the largest piece of content is on screen, often a photo or the heading | up to 2.5 s | over 4 s |
| INP | How quickly the page responds when you tap, click or type | up to 0.2 s | over 0.5 s |
| CLS | How much the page shifts while you are looking at it | up to 0.1 | over 0.25 |
The LCP is the metric that comes closest to “how fast does my site load”. That is what the rest of this article is about. More about all three metrics in Core Web Vitals in 2026.
What does a slow site cost?
A slow site costs you in three places. Visitors drop off before they have seen anything. If you pay for ads, you also pay for the click of someone who leaves before the page appears. And for search, Google writes that it uses the Core Web Vitals in its ranking systems, and recommends getting them in good shape. At the same time Google writes that it always wants to show the most relevant content, even when the page offers a less pleasant experience.
Picture someone with a leak in the kitchen, looking for a plumber. They open three results at once. Whoever shows a phone number and a clear button first gets the call. The other two may load properly a second later, but by then the choice has been made.
How much revenue a second costs exactly differs per site and per visitor. We have not measured that ourselves, so we do not give a fixed number. What we can say is that speed is largely in your own hands, unlike many other things that count.
How fast are the sites we built?
We measure every site with the same method: a phone screen, a processor four times slower, a slow connection and an empty cache, without cookies or consent. We measure each homepage three times and take the middle result. We measure the time until the largest piece of content is on screen (the LCP). The table lists the sites whose case page states a load time.
| Site | Built with | Load time on a phone | Measured on |
|---|---|---|---|
| Basestep (our own product) | Astro | 0.58 s | 24 October 2026 |
| snelrepareren.nl (our own product) | Own server in Node.js, pages as plain HTML | 0.74 s | 24 October 2026 |
| TaalPro | Astro | 0.78 s | 22 October 2026 |
| Atiq Tandartsen | Astro | 0.82 s | 3 November 2026 |
| Daily Phones | WordPress | 0.94 s | 22 October 2026 |
| Ikiz | Astro | 0.98 s | 22 October 2026 |
| Sittard Taxi | Nuxt (homepage with booking wizard) | 1.23 s | 22 October 2026 |
| Taxi Grou | Astro | 1.63 s | 22 October 2026 |
| ArtsenPool (not yet launched) | Next.js | 1.7 s | 24 October 2026 |
In this measurement all nine stay below the 2.5 seconds that Google calls good. Note that for that threshold Google looks at real visitors, while we measured in a test setup. A phone on a faster connection loads the same site more quickly.
Why is one site faster than another?
The technology matters, but it does not explain everything. The weight of the page and the number of files also explain the difference only in part. Two examples, with the weight of the homepage as stated on the case page:
- Sittard Taxi has a booking wizard in Nuxt on its homepage: 412 KB in 15 requests. The service and town pages are static pages of around 32 KB of HTML.
- Taxi Grou loads 355 KB in 8 requests, with a request form on the homepage.
So Taxi Grou loads fewer files and is still slower. Often a single element at the top, such as a large photo or a form, decides the load time. A cookie notice can also be the largest element on the screen, and then it sets the LCP instead of your own content. Note: the weights on the case pages were measured on 24 September 2026, the load times of these two sites on 22 October.
WordPress is not necessarily slow. Daily Phones runs on WordPress and was on screen after 0.94 s in our measurement. Whether a WordPress site is fast depends on the theme, the plugins and the images.
Where does the delay usually come from?
Slow sites often share the same causes:
- Images that are uploaded and served far larger than they need to be.
- Scripts and fonts that hold up rendering of the page.
- A pile-up of trackers, plugins, chat buttons and marketing tools, each with its own files.
- A booking module, map or video from another party that already loads when the page opens.
- Hosting that does not match the traffic, or is set up wrongly.
- No caching, so every visitor fetches everything again.
How do you measure the speed of your own site?
You do not need a specialist for it. This is how you get a fair picture:
- Open Google's PageSpeed Insights and enter your address. You get numbers from real visitors, if there are enough of them, and a lab test.
- Look in Search Console at the Core Web Vitals report. It groups your pages and shows whether each group is good, needs improvement or is poor, based on real visitors.
- Measure on a phone, not just on your laptop. Or use the developer tools in Chrome to slow down the connection and the processor.
- Measure three times and take the middle result. A single measurement can be fast or slow by chance.
- Measure more than the homepage. Include an ordinary page too, such as a service or an article.
- Write down the date. That way you can see later whether a new feature or photo made the site slower.
How do you read the result?
PageSpeed Insights gives two kinds of results. At the top, if there are enough visitors, you see the numbers from real visitors over the past 28 days; if the page has too little traffic, you see the numbers for your whole site. Below that is a lab test with a score from 0 to 100. That score summarises one test run and is useful for finding problems. For the Core Web Vitals, Google looks at real visitors. So a score of 70 with good numbers from real visitors is no reason to panic, and a high score with poor numbers from real visitors is a reason to look further.
If the LCP is too high, first check which element is the largest. Usually it is a photo or a heading in a custom font. Load time and the Core Web Vitals are also part of our technical SEO audit.
Why measure on a phone?
A laptop on office wifi is the most favourable situation there is. Your visitor is often on a phone, with a slower processor and a connection that comes and goes. That is also why Google keeps phone and desktop apart. A site that appears instantly on your laptop can stay blank for seconds longer on a phone with a slow connection. So measure in your visitor's situation, not in your own.
What can you do about a slow site yourself?
Not everything needs a developer. You can often take these steps yourself, or ask for them specifically:
- Make your photos smaller. A photo from your phone is often much larger than the screen will ever show. Resize photos to the size at which they are displayed, in a modern format.
- Clean up scripts. Go through the trackers, pixels and tools on your site. Whatever nobody uses any more can go.
- Load heavy parts only when they are needed. A map, video or chat button does not have to load before someone clicks on it.
- Limit fonts. Every extra font and every extra weight is one more file.
- Look at your hosting. If your site sits on a server that answers slowly, the rest only helps in part.
How do we build speed in?
Fixing speed afterwards is like mopping the floor with the tap still running. That is why it sits in the foundation for us: our sites in Astro are static, and our applications in Nuxt build the page on the server, so the visitor gets a ready-made page straight away instead of a blank screen that still has to fill up.
- Images in modern formats, at the right size and only loaded when they come into view.
- Loading only what the page really needs.
- Fonts loaded carefully, without text that suddenly changes shape.
- Caching set up from day one.
- A performance budget that runs alongside the build from the start.
What should you ask your developer?
If you are having a site built, ask about speed beforehand, not afterwards:
- How fast does a comparable site you built load, measured on a phone, and on what date?
- Which scripts from other parties will be on it, and why?
- How do you deal with photos we add ourselves later?
- Do you measure the speed before handover, and do we get to see the result?
- Who makes sure the site stays fast after launch?
How do you keep a fast site fast?
A fast site gets slow again on its own if nobody keeps an eye on it: every new feature, tracker or photo nibbles away at it. That is why measuring is part of maintenance. In our maintenance plan we check periodically how fast your site loads and whether images that are too heavy have been added. Small things we fix straight away.
When this does not apply
Our numbers are our own measurements of the homepage, in a test setup, on one day per site (22 October, 24 October or 3 November 2026). They are not data from real visitors in Google, and they are not the same as the result from PageSpeed Insights, which measures differently. The table only lists the sites whose case page states a load time.
What we do not know: we have not measured how many extra requests or how much extra revenue a faster site brings in. Our test setup does not measure how quickly a page responds to a tap (INP), and we only see shifting during loading. And a fast homepage says nothing about a page with a heavy map or video; measure that separately.
Frequently asked questions
How fast should a website load?+
What is a good LCP?+
Does my site's speed affect where I rank in Google?+
Is WordPress always slower?+
How do I measure the speed of my site myself?+
Want to know how your own site does? Send us your site's address and we will tell you where the time goes. More about how we build in web development. You work directly with Nawid, with a fixed team of five specialists.
Sources
Consulted on 3 November 2026.
- web.dev, Web Vitals (thresholds and the 75th percentile)
- web.dev, Largest Contentful Paint (LCP)
- web.dev, Interaction to Next Paint (INP)
- web.dev, Cumulative Layout Shift (CLS)
- Google Search Central, Understanding Core Web Vitals and Google search results
- Google Search Central, Understanding page experience in Google Search results
- Google, About PageSpeed Insights (28 days, data for the whole site)
- Google, Search Console Help, Core Web Vitals report
- Empireyo, Process (performance budget) and Web development (measurement before launch)
- Empireyo, cases Basestep, snelrepareren.nl, TaalPro, Atiq Tandartsen, Daily Phones, Ikiz, Sittard Taxi, Taxi Grou and ArtsenPool
Written with the help of AI, reviewed and edited by Nawid Rasa, who is responsible for the content. Sources are listed below the article. Spotted an error? Email hi@empireyo.com.
Ready for a platform that performs?
We measure where your site stands now and show what can be gained technically, in concrete terms and without jargon.
Your request has reached us. You will usually hear back within two working days. Prefer direct contact? Email hi@empireyo.com or message us on WhatsApp with the chat button at the bottom right.