We Benchmarked Our Own Site and It Was Humbling

We build shirts about performance culture. It would be a little rich to do that and never actually check our own numbers, so we ran our own site through the same tools we joke about on our merch. Real Lighthouse audit, real Core Web Vitals, no cherry-picking the run that happened to land well.

Our Largest Contentful Paint was fine. Not the kind of “fine” you brag about — the kind of “fine” where you glance at the number, nod, and quietly move your cursor to the next metric. Our Time To First Byte was solid, which we’ll take a small amount of pride in.

The genuinely humbling part was Cumulative Layout Shift. We have a shirt that says “Reduce Your CLS, Not Your Ambition,” and our own storefront had a small layout shift caused by a product image loading in without a reserved space above it. We found this out the same way most people find these things — by looking, on purpose, instead of assuming it was fine because nobody had complained yet.

We fixed it the boring way: explicit width and height attributes, a reserved aspect ratio, no clever tricks. There’s no plot twist where the bug turns out to be an ancient database query nobody’s touched since 2019. It was just a missing dimension attribute.

This is also, genuinely, a big part of why the site is built the way it is: it doubles as a benchmark environment for web-performance tooling, so different setups and optimizations get tested against the same real storefront over time — not a synthetic demo page, an actual site people buy shirts from.

We’re not going to pretend this makes us performance royalty. But there’s something clarifying about being the exact audience your own jokes are aimed at, and then getting mildly dunked on by your own tooling. CLS is fixed. LCP is fine. TTFB is the one thing we’re currently smug about, and we’re allowed exactly one of those per audit.