top of page
Search

Website Speed Optimisation: A Practical Guide for Site Owners

Oct 25
9 min read

Site owner checking website performance

To speed up a website, measure it with real-user data first, then fix the biggest causes in order: the server response, the hero image, render-blocking code and heavy scripts. Most sites gain more from five or six of these fixes than from dozens of small tweaks. This guide sets out the targets Google uses, the fixes that matter most, and the traps that make a site slower or break it.



TL;DR

 

  • Aim for the three Core Web Vitals “good” thresholds at the 75th percentile of real visits: LCP of 2.5 seconds or less, INP of 200 milliseconds or less, CLS of 0.1 or less.

  • Measure with field data (real users) as well as lab tests. Lighthouse cannot measure INP properly.

  • Fix in order of impact: server response and caching, images, render-blocking code, then third-party scripts. Leave the small jobs until last.

  • Never lazy-load your main hero image, and never switch on every cache layer without testing.

  • On WordPress, most slowness comes from hosting, uncached pages, oversized images and too many plugins.

 

Table of Contents

 

 

What “fast” means in numbers

 

Speed is not one number. Google’s Core Web Vitals split it into three, and a page counts as good only when all three pass. According to web.dev’s Web Vitals guidance, each is judged at the 75th percentile of field data:

 

Metric

What it measures

Good

Poor

Largest Contentful Paint (LCP)

How quickly the main content appears

2.5 s or less

over 4.0 s

Interaction to Next Paint (INP)

How quickly the page reacts to taps and clicks

200 ms or less

over 500 ms

Cumulative Layout Shift (CLS)

How much the layout jumps around

0.1 or less

over 0.25

Anything between good and poor is “needs improvement”. The 75th percentile matters: your site passes only if three in four real visits meet the threshold, so a fast office connection tells you very little.

 

Time to First Byte (TTFB) is not a Core Web Vital, but it underpins the rest. The GDS Way, the UK government’s service standards site, is the source for the guidance on TTFB: aim for under 800 ms, and treat anything over 1.8 seconds as poor.

 

One recent change is worth knowing. As of December 2025, LCP and INP are supported across all major browser engines, Safari included. Site owners no longer have a blind spot for iPhone visitors, and monitoring should not rely on Chrome data alone.

 

Measure before you change anything

 

Guessing wastes time. Spend an hour on diagnosis and the rest of the work gets shorter.

 

Field data and lab data answer different questions

 

Field data comes from real visitors. Lab data comes from a simulated test, such as Lighthouse, run in a controlled setting. The two are not interchangeable. Google’s guide to measuring Web Vitals recommends collecting both. Lab tools cannot measure true INP because they do not simulate real interactions, so Total Blocking Time is used as a lab proxy, while real-user data remains the only accurate view of responsiveness.

 

In practice:

 

  1. Run PageSpeed Insights on your home page, your top service page and your slowest template. Look at the real-user section first, if your site has enough traffic to show one.

  2. Note which of the three metrics fails. Each tends to fail for its own reasons, as this Core Web Vitals guide explains: slow LCP usually traces to a large hero element, while the other two have different causes.

  3. Test on a mid-range phone on mobile data, not only on your laptop. Most visitors are not on your connection.

  4. Write the baseline numbers down. You will need them to prove the work paid off.

 

Fix things in order of impact

 

The GDS Way advises prioritising performance tasks by impact. High-priority jobs include placing styles at the top of the page, deferring non-critical scripts, compressing static files with Gzip or Brotli, and minifying CSS and JavaScript. Low-priority jobs, such as trimming cookie size or adding service workers, come after those foundations. Here is how that works out on a typical small-business site.


High and low priority website speed fixes

1. Server response and hosting

 

If the server takes a second to answer, nothing you do in the browser can make up for it. According to the same GDS guidance, a high TTFB is often down to underpowered hosting or slow database queries in dynamic systems like WordPress. The remedies are server-level full-page caching and an edge CDN that serves assets closer to the visitor.

 

For a business selling to customers in Surrey, Hampshire or London, check that your host has UK or nearby European servers. Cheap shared plans that serve hundreds of sites from one machine are a common cause of slow response at busy times.

 

2. Images

 

Images are usually the heaviest part of a page. The Page Speed Optimisation guide on SEO Handbook lists large, uncompressed JPEG and PNG files among the most common causes of slow pages.

 

What to do:

 

  • Convert to a modern format. WebP typically cuts file size by 25 to 35 per cent against JPEG at similar visual quality, according to industry data. AVIF can compress further. Use the <picture> element so older browsers get a fallback.

  • Resize to the display size. A 4000-pixel camera photo shown in a 800-pixel column is wasted bandwidth.

  • Set width and height attributes on every image so the browser reserves space and the page does not jump. This is the cheapest CLS fix there is.

 

3. The hero image and lazy loading

 

Lazy loading is useful, but not everywhere. Applying it to the image at the top of the page can damage your LCP score because the browser must wait for JavaScript before fetching it. Practitioners exclude hero images from lazy loading and often add fetchpriority="high" to them. Lazy-load everything below the fold, and keep the first image eager.

 

Many WordPress themes and plugins lazy-load every image by default. Check your hero image is not caught by that setting.

 

4. Render-blocking CSS and JavaScript

 

A browser will not draw the page until it has handled the styles and scripts in the head. Three changes help most:

 

  • Inline or prioritise the CSS needed for the first screen, and load the rest later.

  • Defer scripts that are not needed for the first view.

  • Minify and compress. Brotli or Gzip on the server cuts transfer size sharply for text files.

 

INP problems usually come from long JavaScript tasks that keep the page busy. Chat widgets, sliders, tracking tags and page-builder scripts are the usual suspects. Remove what you do not use before trying to optimise what you do.

 

5. Third-party scripts

 

Analytics, ad pixels, booking widgets, embedded maps and video players each add requests and code you do not control. List them, ask what each earns you, and delete those that earn nothing. Load the remainder after the page is usable. A video embed can be replaced by a thumbnail that loads the player only on click.

 

Caching without breaking the site

 

Caching is where well-meant optimisation goes wrong. Performance specialists make the point that the aim is fast, correct, predictable behaviour, not maximum caching. A healthy setup layers several caches:

 

  1. Browser cache for static files such as images, CSS and JavaScript.

  2. CDN edge cache to serve those files near the visitor.

  3. Full-page cache, as static HTML, for anonymous visitors.

  4. Persistent object cache, such as Redis, to reduce database work.

 

Switching on every layer blindly can produce stale content or broken features. Carts, checkouts, account pages and forms must be excluded from full-page caching. After any change, test the contact form, the basket and the login.

 

For files that rarely change, use content hashing. If you add a hash of the file’s contents to its name, such as style.a1b2c3.css, you can set the cache lifetime to a year. When the file changes, so does the name, and visitors fetch the new version straight away with no manual cache clearing.

 

Speeding up a WordPress site

 

WordPress powers a large share of small-business sites, and it has predictable weak points. Work through this list in order:

 

  1. Check the host and PHP version. Slow queries and weak hosting drive up TTFB, as covered above.

  2. Add full-page caching at server level if your host offers it, or through a well-regarded caching plugin. Use one caching plugin, not three.

  3. Audit plugins. Each active plugin can add scripts and database queries. Deactivate anything you are not using, then delete it.

  4. Replace heavy page-builder layouts on key pages if they ship large amounts of JavaScript and nested markup.

  5. Optimise images on upload, with automatic WebP conversion and sensible maximum dimensions.

  6. Clean the database: old revisions, expired transients and spam comments.

  7. Re-test after each change, so you know which one helped and which one broke something.

 

If you run an online shop, speed work should extend to the basket and checkout, where delay costs sales directly. This practical Shopify checkout optimisation write-up shows how performance and extensions affect conversion on that platform, and the principles of auditing what loads at checkout apply to other stores too.

 

Does speed affect search rankings?

 

Core Web Vitals are part of Google’s page experience signals, and Google surfaces them across its tools, per web.dev. Treat them as one factor among many. Relevant content still matters more than a perfect score. The commercial argument is stronger than the ranking one: a page that responds quickly keeps visitors long enough to read it, call you or fill in a form. Slow pages lose those visitors before the content gets a chance.

 

Security belongs in the same conversation. A site served over HTTPS is the baseline for visitors and for search checks, and our post on SSL certificates and HTTPS checks for site owners covers what to verify. Get it right early so that redirects do not add delay.

 

Our view: speed is a design decision, not a final polish

 

We build and maintain websites alongside IT infrastructure, cloud and networking work, and that mix shapes how we see speed. A slow site is rarely one bad plugin. It is usually the sum of decisions made early: the hosting tier, the size of the images the client supplied, the theme, the third-party tags marketing added over the years.

 

So we would tell any owner the same three things. First, set a budget before the build: a target for LCP, INP and CLS, and a limit on the third-party scripts allowed. Second, treat infrastructure as part of the website. Server response, caching and delivery sit under every page, and a front-end fix cannot undo a weak foundation. Third, measure on real devices and keep measuring, because sites get slower as content and tags accumulate.

 

And resist the urge to chase a perfect 100 in a lab tool. A site that passes all three Core Web Vitals for real visitors, loads quickly on a mid-range phone and keeps working when you add content is a better result than a flawless synthetic score on a page nobody can edit.

 

— Andrii

 

Need a fast site built or fixed?

 

If you would rather hand this over, our web site creation service covers building new sites and improving existing ones, backed by our IT infrastructure and cloud expertise. Tell us your current pain points and the pages that matter most to your business, and we will take it from there.

 

A one-week action plan

 

Day

Task

Expected effect

1

Run PageSpeed Insights on three key pages and record LCP, INP, CLS and TTFB

A baseline to beat

2

Check hosting, PHP version and full-page caching

Lower TTFB

3

Resize and convert images to WebP, add width and height

Better LCP and CLS

4

Make sure the hero image is not lazy-loaded; add fetchpriority

Better LCP

5

Remove unused plugins and third-party scripts, defer the rest

Better INP

6

Enable Brotli or Gzip, minify CSS and JavaScript

Smaller transfers

7

Re-test, test forms and checkout, record the new numbers

Proof and safety check

Field data takes time to update, because it reflects real visits over a rolling period. Do not expect the real-user numbers to change overnight even when lab results improve straight away.

 

FAQ

 

How do I make my website run faster?

 

Start with the biggest causes: slow server response, oversized images, render-blocking scripts and unnecessary third-party tags. Measure first so you know which applies to you. Then fix them in order of impact and re-test after each change.

 

What is a good page speed score?

 

Focus on the Core Web Vitals thresholds rather than a single score: LCP of 2.5 seconds or less, INP of 200 milliseconds or less and CLS of 0.1 or less, measured at the 75th percentile of real visits. A high lab score helps, but passing in field data is what counts.

 

Why does my score differ between tools?

 

Lab tools simulate one visit on one device and connection, while field data reflects many real visits. Lab tests also cannot measure INP directly. Use lab data to diagnose and field data to judge.

 

Should I lazy-load all my images?

 

No. Lazy-load images below the fold, but keep the main image at the top of the page loading immediately. Lazy-loading it can hurt your LCP.

 

Sources

 

Recommended

 

 
 
 

Comments


© 2026 Epicloud Operations UK Limited | 5 Brayford Square, London, E1 0SG | VAT:GB456501595 | Terms&Conditions

bottom of page