A slow website adds friction and can cost conversions. The exact effect differs by audience, device and funnel. Optimize against real Core Web Vitals and your own conversion data, not a universal promise.
1. Why sub-second matters now
Core Web Vitals measure loading, responsiveness and visual stability. Google's current guidance defines good LCP as 2.5 seconds or less for at least 75% of visits. Measure mobile and desktop separately because real devices and connections vary widely.
2. Four common bottlenecks
- Too much JavaScript: large bundles require download, parsing and execution time
- Images that are too heavy: missing srcset, unsuitable formats and excessive resolution slow LCP
- Third-party scripts: analytics, widgets and embeds compete with primary content
- Slow origin: server work, database calls and distance increase initial response time
3. The architecture we use
Our default is a small public runtime, cacheable pages and client-side JavaScript only where interaction needs it:
- Astro and static output: HTML first, JavaScript only for real interaction
- Edge delivery and caching: assets close to the visitor and long caching for versioned files
4. Quick wins you can do today
- Modern image formats: use WebP or AVIF with an appropriate JPEG/PNG fallback where needed
- Load fonts deliberately: preload only critical variants and use font-display
- Lazy-load below the fold: reserve dimensions to prevent layout shifts
- Limit third parties: load only what adds measurable value and after consent where required
5. When a rebuild pays off
Baseline traffic, conversion, margin per conversion and Core Web Vitals. Then improve one bottleneck at a time and compare the same KPIs. This turns a rebuild into a measurable business case rather than a generic speed claim.
6. What sub-second does NOT fix
A fast site does not fix a weak offer, unclear positioning or a cumbersome funnel. Speed removes technical friction; offer, proof, copy and usability then determine how many visitors actually convert.