Restructuring Q-Bot's highest-entry webpage around evidence that most visitors already know the brand.
Q-Bot is a UK company offering robotic underfloor insulation: spray foam applied beneath suspended timber floors by a robot, requiring only a few floorboards to be lifted. It serves four core segments: owner-occupiers of suspended timber floors, park home owner-occupiers, landlords and social housing providers, and retrofit contractors, within a regulated retrofit landscape (PAS 2035, ECO4, GBIS, TrustMark, IBG).
The homepage is the site’s highest-entry page, so any change to it carries outsized risk and opportunity. I treated the redesign as a structured experiment rather than a visual refresh: hypothesis first, pre-registered metrics and isolated changes wherever possible.
Traffic-source analysis showed a mismatch. The audience arriving on the homepage is largely brand-aware, with the large majority arriving via direct or branded organic search.But the page was structured as an awareness and education piece, explaining why insulation matters to visitors who had, in most cases, already searched for Q-Bot by name.
Six months of GA4 analysis quantified the cost. Fewer than one in five visitors progressed from the homepage to a key page and only a small fraction reached the Contact page. These visitors were scrolling deep and engaging with the content, but they just were not efficiently routed anywhere.
Exact visitor volumes and detailed funnel figures are confidential and omitted.
Before committing to a hypothesis, I ran the analysis through critical review. High intent is an inference, not a finding: "q-bot" aware traffic proves brand awareness but not purchase readiness. In addition, the single largest leak in the funnel was the drop-off between key pages and Contact, which is outside of this project. So I acknowledged explicitly what the homepage experiment can and cannot be credited with.
Hypothesis: Visitors likely already know Q-Bot and are looking for proof, process clarity and next steps, but are not finding them quickly or clearly enough.
The redesign bundled five changes: a home suitability check, a 25-year insurance backed guarantee disclaimer, a social-proof banner, a process overview and the demotion of low-intent educational content further down the page.
Four simultaneous changes would normally make it impossible to know which one moved the numbers. So the demotion was split out as an independently measurable and reversible step, separately from the additions. Demoting rather than deleting also preserved SEO and AEO value (non-branded search capture, internal linking, AI answer-engine extraction) and kept serving the segments that genuinely need the "why".
Every component was built as a Raw HTML block as imposed by the WP Bakery page builder. That forced a set of defensive conventions: doubled-class selectors to override theme styles without !important, no bare element selectors (they bleed into theme layout), fallback values on every var() call since custom properties may not resolve inside Raw HTML blocks, and accessibility as standard: prefers-reduced-motion blocks, keyboard navigation, semantic headings, and ARIA attributes in every interactive component.
Each component file is duplicated and date-labelled before changes, preserving a rollback path for every iteration.
The new page reorders the journey around proof, process and next steps:
The intention is to enable faster trust-building, clearer value communication and reduced cognitive friction for a largely brand-aware audience. This will be measured by progression from homepage to key pages and homepage engagement.
The new homepage was deployed on July 15th 2026, and the pre-registered measurement window is now running.
The experiment stands on a single pre-registered metric: progression from the homepage to key pages (17.8% before deployment) targeting 24% after, and compared on the same matched Jan-Jun window a year on. The target was derived from the baseline.
Acknowledging the limitations: industry policy changes, marketing activity and competitor moves could influence the results. Results will be updated here.
Next project →Outside of work, I'm currently building side projects and am always happy to discuss or get involved in problem-solving challenges. Have something in mind? Get in touch.
Email me