Skip to content
Klymax
All insights

Your website is slow: what it costs you, and how to measure it

Core Web Vitals in plain English, free testing tools and fixes ranked by their impact.

Klymax9 min read

Illustration of servers embedded in a rocky foundation beneath a web page.

Audio version

0:00

You cannot judge your website's speed from your desk. You know the address, your browser has everything cached and your connection is good. Your visitors arrive on 4G while standing in a queue, using a three-year-old phone.

What a slow website actually costs

There are two effects, and the second matters more than the first. The direct effect is abandonment: after three seconds, a significant share of visitors close the page before they have even seen it. They have not judged your business. They have not seen it.

The indirect effect is on rankings. Google measures the loading experience of real visitors and uses it as a signal. Its importance is modest compared with content relevance, but it can separate two equally relevant pages. In a competitive market, that is exactly where the difference is made.

The three metrics that matter

Google uses three metrics called Core Web Vitals. They may look unremarkable, but they describe precisely what a visitor experiences.

  • LCP: the time until the largest visible element appears, often the image at the top. Aim for under 2.5 seconds.
  • INP: the delay between a click and the page's response. Aim for under 200 milliseconds. This metric replaced FID in March 2024; an audit still discussing FID is working with a retired metric.
  • CLS: layout shifts while the page loads. Aim for under 0.1.

How to measure it for free

Google's PageSpeed Insights gives you all three metrics for any address, on mobile and desktop, with a list of fixes ranked by estimated benefit. It is free and requires no account.

One distinction will help you focus on the right problem: the tool shows two types of data. Lab data comes from a simulated test and is available to everyone. Field data comes from real visitors and only appears when a website has enough traffic. It is the second type that matters to Google.

Another habit to adopt: always test the mobile version. Google indexes the mobile version of a website, and it is almost always the slower of the two.

Illustration of a web page protected by a transparent structure with green edges.
Illustration: website maintenance continues after launch.

Fixes ranked by their impact

1. Images, by far the leading cause

Nine out of ten slow websites are held back by their images. A photo straight from a camera weighs 5 MB and is 6,000 pixels wide, yet appears in a 600-pixel block. Visitors download ten times the width they will actually see.

Three steps are enough. Resize images to their actual display dimensions. Convert them to WebP, which is around 30% smaller than a JPEG at equivalent quality. And defer the loading of images further down the page.

The same photo, with different processing
Straight from the camera, 6,000 px
4,800 to 5,600 kB

What most websites actually send.

Resized to 1,600 px, JPEG
280 to 340 kB

Resizing alone does most of the work.

Resized and converted to WebP
85 to 110 kB

No visible difference, one third of the file size.

85 kBLogarithmic scale5,600 kB

An article hero image. Going from the original to the optimised version cuts the file size by more than fifty times, with no visible difference on screen.

2. Third-party scripts

Every tool added to a website, whether live chat, a cookie banner, an interactive map, an advertising pixel or analytics, adds a request to a server you do not control. If that server is slow, your website is too.

A fast website with six third-party tools piled on top can become slow without a single line of its own code changing. Take stock once a year and remove anything no longer useful. Nobody remembers installing half of them.

Illustration of a mountain built from successive layers in front of a web page mockup.
Illustration: a finished page rests on several stages of design.

3. Accumulated plugins

On WordPress, each plugin loads its own code on every page, including pages where it serves no purpose. A form plugin loaded on all twenty pages when it is only needed on the contact page slows down the other nineteen.

Beyond ten plugins, the question is no longer which one to add, but whether the CMS is still the right tool.

4. Hosting

It is the cause most often blamed and least often responsible. Decent shared hosting handles a business website without difficulty. It becomes the limiting factor when a website has heavy traffic, many dynamic pages or an online shop.

The telltale sign: if server response time consistently exceeds 600 milliseconds, images are no longer the problem.

What a fast website means to a visitor

A fast website impresses nobody. That is the point: nothing gets in the way, people find what they need and call. Speed does not directly win customers. It stops you losing them before you have had a chance.

Start with images. On most websites, that change alone brings all three metrics back into the green, without a redesign or a new hosting provider.

Frequently asked questions

01How can I tell if my website is slow?

PageSpeed Insights is free, requires no account and gives you all three metrics for any address. Test the mobile version: it is the one Google indexes and almost always the slower of the two.

02Does speed really affect Google rankings?

Yes, although its effect is modest compared with content relevance. It can separate two equally relevant pages, making it decisive in a competitive market and less important in an uncontested one.

03What should I fix first?

Images, in nine cases out of ten. Resizing them to their actual display dimensions and converting them to WebP often reduces a page's file size by a factor of five, with no visible difference.

04Do I need to change hosting provider to make my website faster?

Rarely, and never as the first step. Hosting becomes the limiting factor when server response time consistently exceeds 600 milliseconds. Below that, the problem lies elsewhere, usually with images or third-party scripts.

05What is INP, and why has it replaced FID?

INP measures the delay between an interaction and the page's response. It replaced FID in March 2024. Google stopped publishing FID in September 2024. An audit that still mentions it is working with a retired metric.

Keep reading

Related articles

Your turn

What if our next project
were yours?

Imagine your website

Your free mockup, with no obligation.