Oversized images
A 4000px photo shown at 400px, or five hero images loading at once. Usually the single biggest win.
Home/Website Speed Optimization
A perfect lab score means little if the site still takes four seconds on a mid-range Android over 4G. Speed work is not about chasing a number: it is about the first screen appearing quickly, the layout not jumping, and taps responding. I measure on the devices your customers actually use, fix the biggest causes first, then measure again.
Almost every slow site I look at has the same five causes. Fixing them in order is cheaper than rebuilding.
A 4000px photo shown at 400px, or five hero images loading at once. Usually the single biggest win.
Page builders, chat widgets, sliders and tracking tags each add blocking work before the page responds.
Web fonts loading four weights from a third party, with no fallback, blocking the first paint.
Every visitor waits for the server to rebuild the same page. A cache and CDN remove that.
Ads, banners and late-loading images pushing content down as the visitor tries to tap.
Record mobile field data, lab runs and a waterfall of what loads first.
Images, fonts, scripts and caching — in order of impact, not order of convenience.
Re-run the same tests on the same connection profile and compare.
Add lightweight checks so new content does not quietly undo the work.
Real-user Chrome UX data and Search Console Core Web Vitals decide what actually needs fixing.
Lighthouse and WebPageTest on a slow 4G profile — the conditions most visitors are in.
A 100 in a lab tool is not the goal. A fast first screen and a stable layout are.
Speed is a confirmed ranking input, but a small one next to relevance. The bigger wins are usually conversion and bounce rate.
Most small business sites can reach a good mobile score. It depends on how much third-party script you are willing to remove.
Usually not. Images, fonts, scripts and caching fix most of it. A rebuild only makes sense when the template itself is the problem.
Caching plugins help, but they hide causes rather than remove them. I fix the cause and use a cache on top.
The same tests, same device profile, before and after — plus field data once enough real visits have been recorded.
A short message with the site address is enough for a first read on what is slow and what it would take to fix.