Every frontend developer knows the usual performance playbook: compress images, preload the hero, remove render-blocking JavaScript, reduce bundle size. Those optimisations matter, but they stop making a meaningful difference when the server itself needs more than a second to respond. At that point, the bottleneck has moved somewhere deeper in the stack.
This talk is about what happens after you’ve exhausted the obvious wins. I’ll break down where TTFB actually comes from and how to improve it at the source. That means looking beyond frontend payloads into framework-level caching with use cache and ISR, edge rendering strategies, HTTP/3, 103 Early Hints, and using Server-Timing to measure what’s really happening instead of relying on assumptions. But reducing TTFB is only part of the story. In many cases, the better approach is making sure the user never experiences the delay in the first place. We’ll explore how Speculation Rules allow browsers to prefetch or fully prerender pages before navigation happens, turning what would normally be a 1.2-second wait into a transition that feels instand. Alongside that, we’ll look at A/B test data from live e-commerce environments showing the impact these techniques can have on user behaviour and conversion.
We’ll also dive into the back-forward cache, one of the fastest navigation mechanisms available in modern browsers. When it works, pages are restored instantly from memory. Expect live coding and showing real production data.