Overview
Siteimprove’s website needed a clearer, more scalable foundation for people and AI. Using its own Siteimprove.ai platform, the team rebuilt the site around accessibility, performance, streamlined navigation, and agentic search readiness, creating a faster, more consistent digital experience.
The Problem
Siteimprove approached its website rebuild from an unusually transparent position: the company used its own Siteimprove.ai platform to solve the same challenges thousands of customers face every day. They treated their website as ground zero for accessibility, performance, and agentic search readiness, applying the newest capabilities of Siteimprove.ai to their own digital estate.
As Elizabeth Irvine, Senior Director of Growth Marketing at Siteimprove, explains, “This redesign wasn’t just about updating visuals. It was about proving that the way we advise customers is the way we operate ourselves.”
At issue were problems common to organizations everywhere. Navigation had grown dense, with too many links and too little clarity on where users should go. Megan Moeller, Front‑End UX Developer at Siteimprove, saw this firsthand: “We had a lot of links and a lot of content on our nav and on our site, but it was hard to navigate where to go to find the information.”
Under the hood, the technical foundation needed modernization. The shift toward agentic search meant AI systems were interpreting content differently than traditional SEO crawlers. The old site wasn’t built for that reality. Sami Khan, Senior Full Stack Software Engineer, put it plainly: “AI agents don’t come and look at the content first. They go after specific data elements in the code, and we had to make sure those were there.”
Performance gaps compounded the challenge. Mobile speed, image weight, and inconsistent page structures were affecting both user experience and AI discoverability. And because the previous redesign had been a re‑skin rather than a rebuild, the team was still working around legacy templates that made consistency difficult.
The result: a site that looked great at a glance but wasn’t telling the full platform story, wasn’t optimized for AI‑driven discovery, and wasn’t giving users, or internal teams, the clarity they needed.
“This redesign wasn’t just about updating visuals. It was about proving that the way we advise customers is the way we operate ourselves.”
The Solve
The team responded with a full rebuild grounded in accessibility‑first design, agentic search readiness, and a unified content and UX strategy. They partnered with an agency to establish the visual foundation, then took ownership of the implementation, refinement, and rollout.
A major shift was structural. The team streamlined templates, consolidated redundant pages, and rebuilt navigation around clarity and intent. Zach Koss, Creative Manager, describes the shift: “This version is more geared toward what our users want. Content is easily digestible, the flow is logical, and we’re not overwhelming people with too much information.
Optimizely played a central role in making the rebuild scalable. The team used Optimizely’s modular architecture to create repeatable blocks and components that made page building faster and more consistent. As Megan notes, “Receiving the agency’s work as HTML, CSS, and JavaScript meant we could implement it cleanly in Optimizely and use it as a strong reference point for new sections.”
Accessibility became a non‑negotiable pillar. The agency’s designs were required to meet WCAG standards before handoff, and Megan layered in multiple rounds of testing using Siteimprove.ai. “Accessibility was a priority throughout the implementation,” she says, “We used the Accessibility Checker, the Accessibility dashboard, and deeper expert reviews to make sure the site works for everybody.”
On the engineering side, the team rebuilt the technical foundation to support agentic search. They added structured data elements, optimized image formats, and used Siteimprove.ai’s performance insights to identify and fix issues early. “Our performance tools showed us which pages had bad speed and why,” says Sami. “We could tweak, compress, and restructure to get a good score, especially on mobile.”
Siteimprove.ai became the connective tissue across disciplines. Pre‑publish checks caught accessibility, SEO, and QA issues before launch. Weekly scans highlighted emerging problems. Policy helped identify legacy video hosting. The Figma plugin flagged design‑stage accessibility concerns. And DCI scores gave the team a measurable way to validate progress.
The result is a site that is visually unified, structurally consistent, and built for both humans and AI. Comparing the two weeks before and after launch, average time on site rose from 0:49 to 1:13, and pages per session ticked up from 1.39 to 1.42. Average weekly hot MQL creation (demo requests and Digital Insights Report) grew from 52 to 55, and page load time improved from 8.02 to 7.79 seconds, with the team continuing to work toward its goal of under four. As Sami summarizes, “We were lucky that we had a whole suite of tools and validation frameworks at our disposal to make the process smooth and efficient for us.”
