Chasing a Perfect Lighthouse Score on a Real Business Site

Run Lighthouse against this website — the live one, waltbuilds.com, not a stripped demo — and you’ll get 99 or 100 on performance depending on the run, and straight 100s on accessibility, best practices, and SEO. I re-measure before I quote those numbers, because a performance claim you haven’t verified on production is just marketing.

I want to walk through what that took, because the answer is unfashionable: almost everything that made this site fast was a removal. And then I want to be honest about what a perfect score is actually worth, because agencies wave Lighthouse numbers around like they’re revenue, and they’re not — they’re a floor.

Fast is mostly the absence of things

The web industry sells performance as an optimization discipline — audits, tweaks, clever caching. For a business site, that’s backwards. You don’t optimize your way to fast. You get fast by not shipping slow things in the first place.

No JavaScript framework in the browser. This site is built with Astro, which renders everything to plain HTML ahead of time. The pages you’re reading shipped as finished documents — no framework booting up in your browser, no loading spinner while scripts assemble a page that could have just arrived assembled. Most business sites ship hundreds of kilobytes of JavaScript to display what is, functionally, a brochure. The fastest script is the one that doesn’t exist.

No stock photos, no sliders, no video hero. The heaviest things on a typical small-business site are decorative — a 2 MB hero image of a handshake, a carousel nobody swipes. We use real screenshots of real work, compressed hard, and typography doing the heavy lifting. Design restraint is performance work; every visual element earns its bytes or gets cut.

Less copy. This one surprises people. We cut this site’s text roughly in half during build — not for load time, for respect. But cut copy is also cut weight, cut layout work, cut time-to-the-point. The fastest paragraph is also the one that doesn’t exist.

One font, self-hosted. One variable font file served from our own domain. No font service, no third-party request, no flash of invisible text. Nobody has ever left a website because it had one typeface.

The pattern: performance wasn’t a phase of the project. Every add-this decision was also a keep-it-fast decision. That’s only possible when the person designing and the person building are the same people — you can’t bolt this on after a designer hands you a 14 MB Figma dream.

Why a business owner should care at all

Not because of the score. Because of what the score is a proxy for:

Google is grading you on it. Page experience is a ranking signal. It won’t outrank better content, but between two local businesses with similar sites, the one that loads in a second on a phone has the edge — and local search is exactly where those margins decide who gets the call.

Your real visitors are on bad connections. Lab scores simulate a mid-range phone on throttled mobile data. Around here, that’s not a simulation — that’s a customer on rural LTE. A site that scores 60 in the lab is a site that takes eight real seconds for that customer, and most of them are gone by three.

Accessibility 100 means the site works for everyone. Screen readers, keyboard navigation, contrast that’s readable in a truck at noon. That’s customers, not compliance theater.

The honest caveat

Now the part the case studies leave out: a perfect Lighthouse score doesn’t sell anything. It’s necessary, not sufficient. A fast site with confusing copy loses to a slower one that answers the customer’s question. Speed just makes sure your answer arrives — it can’t make the answer good. I’ve seen 100-scoring sites that couldn’t convert a visitor who arrived credit card in hand, and I’ve watched businesses obsess over a 97-to-100 gap that no customer on earth will ever feel.

So here’s the priority order I’d actually give a business owner: say the right thing, to the right visitor, instantly. The first two are strategy and copy. The third is engineering, and it’s the only one Lighthouse can see.

We treat the scores as a shipping standard, not an achievement — every site we build goes out the same way, measured on production before we call it done. The scores aren’t the point. They’re the receipt.