How we sped up page loading by 1.8 seconds for a SaaS from Gliwice
A technology company from Gliwice, dealing with cloud ERP systems, struggled with low conversion on mobile devices. Their page took almost 4 seconds to load, discouraging potential business clients looking for fast solutions. At Silesia Performance Lab, we approached the problem without fluff, focusing exclusively on critical rendering path optimization and server performance.
Bottleneck Analysis: Where were the seconds escaping?
In July 2024, we began a technical audit for the CloudArch Systems platform from Gliwice. The first measurements were ruthless. The Largest Contentful Paint (LCP) was 3.84 seconds on a 4G connection. This is 1.3 seconds above the threshold that Google considers acceptable. The data speaks for itself – every second of delay over three seconds risks losing 32% of mobile users before the content even fully loads.
Our analytical tools showed that the main cause of the slowdown was 56 unnecessary HTTP requests generated by old analytics scripts and a poorly configured chat. The home page weighed as much as 4.2 MB, most of which were uncompressed PNG graphics. We check facts, we don't promise miracles, so we started by eliminating the unnecessary ballast that was blocking the browser for the first 1.2 seconds of the session.
Code Cleanup and JavaScript Script Optimization
Instead of general fixes, we performed a surgical cut in the code. We removed 14 unused libraries left over from old marketing tests. We implemented the Tree Shaking technique to load only those code fragments actually needed to display the home page. This reduced the weight of JS files by 412 KB, which in mobile conditions means a gain of about 0.4 seconds just on the phone's processor processing the code.
We also introduced lazy loading for all 32 product images. Thanks to this, the browser downloads graphics only when the user scrolls near them. The most important CSS styles, responsible for the appearance of the top of the page (Above the Fold), were moved directly into the HTML code. This allowed the user to see readable text just 0.9 seconds after clicking the link.
Shortening the loading time by 1.8 seconds translated into 19% more leads in just 22 days after implementing the fixes.
Server and SQL Database Optimization
The problem wasn't just the front-end. CloudArch's server, located in a Warsaw data center, had a response time (TTFB) of 840ms. For a professional SaaS, this is an unacceptable result. We updated the environment from PHP 7.4 to version 8.2, which in itself improved script execution performance by about 18%. The data speaks for itself – a modern technology stack is the foundation of a stable business.
Additionally, we optimized the 11 most taxing SQL database queries responsible for generating pricing tables. We implemented Redis as an object caching system, allowing the server to serve ready results instead of recalculating them for every client visit. The result? Server response time dropped from 840ms to 152ms. A specific audit brought specific performance gains that every user felt.
Results after 22 Days: Green Light from Google
We finished implementing all fixes on August 14, 2024. Just 6 days later, Google's robots updated the data in Search Console. The LCP indicator stabilized at 1.9 seconds, giving us the long-awaited green status in the Core Web Vitals report. We do technical SEO that works because we rely on parameters that have a direct impact on how Google perceives a site's technical quality.
CloudArch Systems' visibility for 14 key business phrases increased by an average of 24 positions in less than a month. Most importantly, the sales department recorded a 19% increase in demo requests. Customers stopped abandoning the offer before it loaded. At Silesia Performance Lab, we believe the data speaks for itself, and technical optimization is the shortest path to more leads.
