Google’s own web.dev case studies, published directly by the teams that build the Core Web Vitals framework, keep landing on the same conclusion from different companies entirely: page speed predicts conversion more reliably, and more cheaply, than almost anything else measured. Nuvemshop improved image loading performance across a cohort of ecommerce stores and saw an 8.9 percent increase in mobile conversion. Rakuten 24 isolated page speed improvements in an A/B test and measured a 33 percent increase in conversion rate and a 53 percent increase in revenue per visitor. Mail.ru improved layout stability on their homepage and saw conversion rates in core sections rise more than 10 percent. Renault analysed over 10 million visits across 33 countries and found a strong, consistent correlation between faster loading and both lower bounce rates and more completed lead forms.
None of these companies redesigned anything. They fixed measurable technical performance, and the business results were larger than what most visual redesigns report.
Google’s own thresholds for what counts as good performance are specific enough to actually test against. Largest Contentful Paint, roughly how long it takes the main content of a page to visibly appear, should land under 2.5 seconds. Interaction to Next Paint, how quickly a page responds to a click or tap, should stay under 200 milliseconds. Sites clearing both bars at the 75th percentile of real user visits are the ones showing up in these case studies, and Google’s own analysis of achievability across the web found that 2.5 seconds is a realistic target for well-built sites, while faster thresholds like 1.5 or 2 seconds aren’t consistently reachable even for strong performers.
What predicts underperformance, based on the case studies rather than assumption, isn’t usually the visual design at all. It’s whether the technical foundation clears these two thresholds. A beautifully designed page that takes four seconds to show its main content is competing with the seven out of ten shoppers Baymard Institute’s checkout research finds abandon a purchase at the first sign of friction, and slow loading is friction before a visitor has even reached the parts a redesign would touch.
The audit we run before quoting any project starts here rather than with the visual brief, because it tells us something the brief can’t: whether the site’s actual problem is how it looks or how it performs, and those require completely different work to fix. A site that fails Core Web Vitals with a beautiful design still fails. A site that passes with a plain one still converts. Speed isn’t a nice-to-have sitting next to design. In every case study we could find, it outperformed it.
Speed isn't a nice-to-have sitting next to design. In every case study we could find, it outperformed it.