We Tested a Flashy Homepage. Here's Why We Didn't Ship It.
We built a version of this homepage with a live particle canvas, stacked parallax sections, and glass-blur cards. It scored 91 Performance against the live site's 100. Toning the effects down to run once on load instead of continuously recovered nothing on Performance and introduced a new accessibility problem — one that a follow-up fix didn't actually resolve.
The surprise wasn't the cost. It was how little "toning it down" helped — or fixed.
We expected converting continuous effects to one-time load effects to close most of the gap. It didn't move Performance at all. Then we tried to fix the accessibility regression it introduced (adding aria-hidden and prefers-reduced-motion support) and re-tested. Still 96, not 100. We don't know yet what the actual cause is, and we're saying so instead of guessing again.
We kept the plain version anyway
This studio's whole case rests on a live, re-runnable 100/100/100/100. Trading any of it for a more decorative homepage isn't a trade we were willing to make, even for a genuinely fun set of effects.
Want the full numbers, including the SEO-score caveat?
Read the comparison