Marketing sites fail quietly. This one loaded slowly enough to shed impatient visitors, needed a developer for every copy tweak, and — the newer problem — wasn’t showing up when buyers asked an AI assistant to compare tools in its category. None of these were emergencies on their own. Together they meant the site was underperforming as the company’s most-visited sales asset.
Fast by construction
We rebuilt the site static-first on Astro. Pre-rendered pages served from the edge meant there was no server work between a click and the content, and shipping almost no JavaScript kept the site responsive on any device. On the key pages this reached a perfect Lighthouse performance score and a sub-second Largest Contentful Paint — not as a launch-day showpiece, but as the steady state.
Built to be cited, not just ranked
The newer requirement was visibility in AI answers. As we’ve written about, getting cited by an answer engine is mostly good engineering: content reachable without JavaScript, clean semantic HTML, structured data, and answers stated plainly up front. We built the site that way by default, which made it legible to AI answer engines and classic search alike. Ongoing growth of that visibility is the kind of work our sister agency Ripple18 runs.
Finally, we moved content into a headless CMS the marketing team owns, and integrated an AI content assistant to speed up drafting — so publishing stopped queueing behind engineering.
As with our other case studies, the figures are illustrative of the outcome this approach delivers and the client is not named. The method is what carries across: build fast by default, structure content so machines and people can both read it, and hand the keys to the team that lives in the site every day.
If your most important sales page is slow, hard to update, or missing from AI answers, that’s precisely what we build. Let’s talk.