Guide · Tech for owners

Website speed: how to make your site fast

A fast website shows its main content within 2.5 seconds, responds to clicks within 200 milliseconds and does not jump around while loading. Those are Google’s three Core Web Vitals: LCP, INP and CLS. This guide gives you the explanation without jargon, a free way to test your site, the fixes that do the most, and a worked example of what a slow site costs.

11 min read · Updated 1 October 2026

The marketing lead has just launched a campaign on Google and Meta and pays for every click. The landing page looks great on the office’s big screen, but on a phone on the train it takes five seconds before the top image appears, and the button jumps down just as you try to tap it. Some of the paid visitors are gone before the page has finished. Speed is one of the few things on a website that affects Google, user experience and sales all at once, and fortunately it is also one of the most measurable.

How fast should a website be?

Google has given a concrete answer. A page counts as fast when at least 75 % of visits meet these three thresholds, measured on real users:

LCP

At most 2.5 seconds until the main content is visible

INP

At most 200 milliseconds for the page to respond to input

CLS

At most 0.1, so the content stays put

What are Core Web Vitals? LCP, INP and CLS explained

Core Web Vitals are three measurements Google uses to describe what it feels like to use a page. They measure what the user experiences: when do I see something, when can I use it, and does it stay still.

LCP: when does the user see the main content?

Largest Contentful Paint measures the time from the user clicking to the largest element in the visible window being shown. It is usually the big image at the top, a video or a headline. The threshold for good is 2.5 seconds. The classic culprits are an image of several megabytes, a slow server, or fonts and scripts that must load before anything may appear.

INP: how quickly does the page respond to a click?

Interaction to Next Paint replaced the older FID metric in March 2024. INP measures the time from the user tapping a button, opening a menu or typing in a field to the screen showing a response. The threshold is 200 milliseconds. Pages with a lot of JavaScript, many tracking scripts or heavy chat widgets often struggle here, because the phone is busy running code when the user taps.

CLS: does the content jump around?

Cumulative Layout Shift measures how much content moves unexpectedly while the page loads. It is the annoying experience of being about to tap a link when an image or an ad suddenly pushes everything down. The threshold is 0.1. Typical causes are images without a set size, banners inserted at the top, and fonts that change size once they have loaded.

Google’s thresholds for the three Core Web Vitals, measured at 75 % of visits.

MetricGoodNeeds improvementPoorTypical cause of problems
LCP≤ 2.5 s2.5–4 s> 4 sLarge images, slow server
INP≤ 200 ms200–500 ms> 500 msToo much JavaScript, third-party scripts
CLS≤ 0.10.1–0.25> 0.25Images without dimensions, late banners

How do you test your website speed for free?

  1. Open Google’s PageSpeed Insights and enter the address of your homepage and your most important landing page.
  2. Choose the Mobile tab and read the field data at the top: does it say "Passed" for Core Web Vitals?
  3. Scroll down to the simulated test and note the score and the three biggest recommendations.
  4. Open Google Search Console and go to the Core Web Vitals report to see which groups of pages have problems.
  5. Test on an ordinary phone on mobile data, beyond the office wifi, and notice whether anything jumps.
  6. Repeat monthly and after every major change, so you catch regressions early.

Lab data (simulated test)

  • One test on a simulated phone and connection
  • Shows the result immediately
  • Good for finding causes and trying fixes
  • Varies from test to test

Field data (real users)

  • Measured on real Chrome users over 28 days
  • What Google uses in its assessment
  • Requires a certain amount of traffic
  • Responds slowly to improvements
Two kinds of numbers that are often confused.

What makes a website slow?

When we review slow sites, the same explanations come up almost every time. Typically it is many small things layered on top of each other over the years, as new tools and campaigns were added:

  • Images uploaded straight from the camera or image bank at full size.
  • Tracking scripts from five different tools, two of which are no longer used.
  • A chat widget, a cookie banner and a video player that all load before the content.
  • A WordPress theme with a built-in page builder and 30 plugins, each adding its own files.
  • Many different fonts and weights fetched from external servers.
  • Cheap shared hosting with no caching, where the server takes a second to respond.
  • A slider at the top of the homepage with five large images, only one of which is seen.

How do you make your website faster? The ten fixes that work best

Fixes sorted by how much they typically deliver relative to the effort.

FixImprovesEffort
Compress images and use modern formats such as WebP or AVIFLCPLow
Set width and height on all images and videosCLSLow
Remove tracking scripts and plugins that are not usedINP, LCPLow
Load images further down the page only as they approach the screenLCPLow
Prioritise the hero image so it loads firstLCPLow
Self-host fonts and use fewer weightsLCP, CLSLow to medium
Load chat, video and maps only when the user needs themINP, LCPMedium
Turn on caching and a CDNLCPMedium
Split JavaScript so only what is needed loads on each pageINPMedium to high
Pre-build pages on the server, for example with Next.jsAll threeHigh, typically with a new site

Does speed matter for Google and for sales?

Google has confirmed that Core Web Vitals feed into its assessment of page experience, which is a ranking signal. It is a modest signal compared with how relevant and useful the content is, so a fast page with thin content will lose to a thorough page that is slightly slower. Speed works best as an advantage when the content is already strong, and as a brake when the site is very slow. The biggest effect is on users. Public studies by Google and large webshops have repeatedly shown that faster pages go together with more people staying and more people buying.

A worked example: you buy 10,000 clicks a month at DKK 8 each, so DKK 80,000. If the page is so slow that one in ten visitors gives up before it loads, you pay DKK 8,000 a month for visitors who never saw your offer. That is DKK 96,000 a year, before counting the customers who stayed but bought less because the site felt heavy. By comparison, a targeted speed review with the most important fixes typically costs a fraction of that. The numbers are an example; use your own cost per click and traffic. Speed is closely tied to conversion rate optimisation, and the two should be planned together.

Is WordPress slower than Next.js?

A well-built WordPress site with a light theme, few plugins and good caching can easily pass Core Web Vitals. The problem is that many WordPress sites end up with heavy themes and page builders, and every new plugin adds code. Next.js is built to deliver finished pages from the server, optimise images automatically and send only the JavaScript each page needs. That makes it easier to be fast from the start and to stay fast as the site grows. We have compared the two in Next.js vs WordPress, and a headless CMS can give your editors the same freedom as WordPress on top of a fast Next.js site.

How does a speed project work?

1

1. Measure the baseline

Field data, simulated tests and Search Console for the most important page types.

2

2. Find the biggest culprits

Which images, scripts and templates account for most of the waiting time?

3

3. Take the quick wins

Images, unused scripts and media dimensions are fixed first, often within days.

4

4. Fix the structural problems

Theme, hosting and JavaScript, or a new site if the foundation is too heavy.

5

5. Monitor continuously

Monthly measurement, so a new script or a heavy image does not creep back in.

From measurement to a site that stays fast.

If you are building a new site anyway, make speed a requirement in the proposal: "every page type must pass Core Web Vitals on mobile at launch". It is one of the most concrete ways to hold a supplier to account. See also what else belongs in an SEO-friendly website.

How does Ceptiv build fast websites?

We build in Next.js, React and TypeScript, with image optimisation, pre-built pages and as little third-party code as possible. Speed is tested on mobile before launch, and the fixed monthly plan covers hosting, updates and monitoring, so the site stays fast. If you have an existing site that has become slow, a UX audit can combine speed with a review of the user experience. Our web packages start at DKK 18,000 plus DKK 600 a month; see them under pricing.

Send the address of your site and your main goals, and you get a fixed price within 24 hours for either a speed review or a new, fast site.

Questions about website speed

Why do we get a different PageSpeed score every time we test?
The score at the top of PageSpeed Insights comes from a simulated test running on Google’s servers, and it is affected by the network, the server’s response time and third-party scripts at that exact moment. Swings of five to ten points are normal. The more important numbers are the field data at the top of the page, which show real users’ experience over the last 28 days. Test several times, look at the average, and treat field data as the final measure. If the site has too little traffic for field data, use Search Console and repeated tests.
What is a good PageSpeed score?
In Lighthouse and PageSpeed Insights, 90–100 is green, 50–89 orange and below 50 red. A green mobile score is a good target for most business sites. Remember, though, that the score is a simulation, and that Google uses the three Core Web Vitals from real users in its assessment. A page scoring 85 with green field data is in good shape. Use the score as a tool for finding problems, and use field data to decide whether you have reached the goal.
Does our cookie banner slow the website down?
It can. Many cookie banners load as a heavy script from an external vendor before the rest of the page may show, and the banner itself can become the largest element on screen and count as LCP. Choose a solution that loads lightly and does not block the rest of the page. It also helps if the banner has a fixed space, so it does not push the content around. Remember that the consent rules still apply, however the banner is built technically.
Does switching to better hosting help?
It helps when the server is the bottleneck, for example when the server’s first response takes more than half a second. That is common on cheap shared hosting with WordPress and no caching. If the problem is large images, many scripts or a heavy theme, new hosting changes very little. A CDN, which serves files from a server close to the user, makes a noticeable difference for visitors far from the server. Measure first where the time goes, and spend the money there.
How long before Google sees the improvements?
Field data in Google’s Chrome report is a rolling average over 28 days, so it typically takes up to four weeks before an improvement shows fully in PageSpeed Insights and Search Console. In Search Console you can ask Google to validate a fix, and Google then tracks progress over that period. The simulated test shows the improvement immediately, so it is useful for confirming the fix works while you wait for field data.
Is speed more important on mobile than on desktop?
Yes, in most cases. Google primarily indexes the mobile version of your site, and phones have both weaker processors and less stable connections than an office computer. A page that feels fast on a new laptop can be heavy on a three-year-old Android phone on 4G. So always test on mobile first, ideally on a mid-range phone. If your audience is mainly on desktop, for example in B2B, the desktop numbers matter a lot as well.

Want us to build it for you?

You get a fixed-price proposal within 24 hours.

Dennis Nielsen

Dennis Nielsen

Head of Operations, Ceptiv

Free consultation

One free hour of advice before you start.

Describe your project and I will contact you as soon as possible, so we can schedule a no-obligation meeting. You leave with practical advice on how to get your project off to a good start.

  • Free
  • 1 hour
  • No obligation