Case Study: How We Improved Website Speed by 70%

This is the story of a real project. A B2B SaaS company came to us with a website that was doing almost everything wrong from a performance perspective.

Not because anyone was being careless. Just because performance had never been made a priority and over time the site had accumulated enough problems to make it genuinely painful to use.

Here is exactly what we found and what we did about it.

Where They Started

When the client came to us their Google PageSpeed score on mobile was 31. Their main content was taking 6.2 seconds to appear on screen. Their homepage was making 94 separate requests every time it loaded.

These are serious numbers. A score of 31 means Google is actively pushing this site lower in search results. A 6.2 second load means most mobile visitors were leaving before they ever saw the page.

The client knew the site felt slow. They had no idea the numbers were this bad.

What We Found in the Audit

Before touching anything we ran a full audit using GTmetrix, PageSpeed Insights and Query Monitor inside WordPress.

The problems fell into five clear areas.

The homepage was loading 23 images totalling nearly 15 megabytes. None of them were compressed. None were in WebP format. None had lazy loading.

The site had 34 plugins installed. 11 were inactive and had not been used in over a year. Several of the active ones were loading scripts on every page even when those pages did not need them.

There was no caching plugin installed at all. Every page visit triggered a full rebuild from the database.

The site was on shared hosting. Server response time was averaging 1.8 seconds before a single file had loaded.

Google Tag Manager was loading 11 additional scripts. Several of them were for tools the marketing team had stopped using months ago.

What We Did About It

We worked through everything systematically over three weeks.

Week one. We compressed the entire image library with ShortPixel. Average image size dropped by 74 percent. We converted everything to WebP and enabled lazy loading across the site. We deleted the 11 inactive plugins and removed four more that were duplicating functionality. We used Asset CleanUp to stop scripts loading on pages that did not need them.

Week two. We installed and properly configured WP Rocket. Page caching, browser caching and database optimisation all set up properly. We set up Cloudflare including minification of HTML, CSS and JavaScript at the CDN level. We audited every script loading through Google Tag Manager with the marketing team. Six were removed immediately because those tools were no longer in use. The remaining five were switched to load asynchronously so they stopped blocking the page from rendering.

Week three. We migrated the site from shared hosting to a managed WordPress hosting environment. Server response time dropped from 1.8 seconds to 180 milliseconds the moment the migration was complete.

The Results

Here is where the numbers ended up after three weeks of work.

That is a 70 percent improvement in overall performance score.

Organic search rankings improved within six weeks of the changes going live. Homepage bounce rate dropped by 34 percent. Time on site increased.

What This Shows You

None of the individual changes were complicated. Image compression, caching, removing unused plugins, cleaning up scripts and better hosting. These are not advanced techniques. They are basics done properly.

The problem is that most sites never have anyone go through them systematically. Issues build up gradually and by the time performance is painful enough to act on there are years of problems to work through.

A performance audit does not have to be a big project. Most of the biggest wins come in the first couple of days of focused work.

Want to know where your site stands and what the highest-impact fixes are? CodingBrackets offers performance audits for WordPress and custom web projects. Get in touch.