Today I Learned

Daily reflections on work and learning—what I did and what I learned.

20267

2026-07-24

When the LCP bottleneck is hydration, not the image — and restrained scroll motion

What I learned improving landing performance — when LCP is slow but the image is small, suspect render delay (hydration) and strip client JS; prefer entrance stagger reveal over flashy scroll motion; structured data must be 1:1 with the screen

  • When LCP is slow but TBT is 0 and the image is small, suspect element render delay (= hydration at the client boundary) and strip the client JS from that surface
  • A short viewport-entrance stagger reveal is often the universal answer over flashy scroll-dependent motion (scroll-jacking)