Why was a catalogue full of images loading in ten seconds?
Kaprel’s site was a catalogue of architectural partition finishes and installations. The obvious explanation was the quantity of imagery. The measured problem was how the site was assembled around that media.
Last reviewed
Deleting images would have damaged the catalogue
Kaprel’s previous development team had left the project, and pages took around ten seconds on average, with the worst at fifteen to twenty. Because the site was image-heavy, the initial instinct was to remove media. That would have made the catalogue faster by removing the product evidence customers came to inspect.
The rebuild found a different issue: the frontend had little control over what it loaded, assets were delivered inefficiently, and content relationships did not connect designs to the systems they belonged to. Image volume was visible; the loading architecture was the cause to address.
What changed
The frontend was rebuilt so page assets were deliberate, the content model connected designs to their systems, and media moved to Amazon S3 with CloudFront delivery. Caching and asset tuning came after the structure was corrected. The catalogue kept its media library.
What the result says, and what it does not
Average page load moved from around ten seconds to under one second, with the worst pages also under one second. The result supports a specific conclusion: in this case, deleting images was not necessary to improve loading. It does not prove that every slow site needs a rebuild, or that the same result will occur on another platform.
Read the full Kaprel case study →
How to diagnose your own media-heavy site
- Measure representative pages on real devices and inspect which resources delay visible content.
- Check image dimensions, formats, responsive variants, lazy loading and delivery location.
- Inspect whether the page loads assets it never displays and whether scripts block rendering.
- Review content relationships and templates before treating every image as an isolated file.
- Compare the same pages after each change; keep image quality and catalogue completeness in the success criteria.
If resource delivery is the issue, targeted optimisation may be enough. If the content model and frontend both force inefficient loading, a rebuild can be the more durable fix. The diagnosis should decide.
Media-heavy performance questions
Do images always make a WordPress site slow?
No. Image size, format, responsive delivery, caching, request priority and the page architecture all affect loading. Removing useful product images is not the only option.
When should a slow site be rebuilt instead of optimised?
When the architecture prevents durable improvements or makes routine changes risky, compare the cost of rebuilding with repeated patches. Measure the current bottleneck first.
What does CloudFront do for site images?
A CDN can serve cached assets from locations closer to visitors. It does not automatically fix oversized images, poor page structure or excessive requests.
What should be measured after optimisation?
Repeat the same representative page and device tests, and compare field data when it is available. Include image quality and content retention in the acceptance criteria.
Kaprel: diagnose the architecture before deleting images
The case study records pages moving from around ten seconds average to under one second while keeping the complete media library.
Read the Kaprel measured rebuild →Measure the bottleneck before removing content.
We will inspect what the browser loads and identify whether the issue is the media, the delivery path or the system around it.