Skip to content
Xentro
Developer workspace with multiple screens of code and neon syntax highlighting
Insights

Core Web Vitals and conversion: case studies the spec hides

Renault saw 13% conversion lift from a 1-second LCP improvement; Vodafone saw 8% sales lift from a 31% LCP gain (web.dev case studies).

  • Feb 25, 2026
  • Reza Mahdavi

Photo by Jakub Żerdzicki on Unsplash

Core Web Vitals discussions usually default to SEO ranking. The more useful framing is conversion, and the case studies are public. Renault tied a 1-second improvement in Largest Contentful Paint to a 13% conversion lift and a 14-percentage-point bounce-rate reduction, measured across roughly 10 million page visits (web.dev Renault case study, 2021). Vodafone A/B-tested a 31% LCP improvement with no UX changes and saw an 8% sales increase plus a 15% lead-to-visit uplift (web.dev Vodafone, 2023).

The pattern repeats across very different verticals. Rakuten 24, an ecommerce property, A/B-tested LCP optimisations and reported a 33.13% conversion-rate increase plus a 53.37% lift in revenue per visitor (web.dev Rakuten, 2023). Tokopedia's optimisation work, which combined 88% script-size reduction with image optimisation and a Svelte rebuild of the homepage, lifted conversion 8% and click-through 35% (web.dev Tokopedia, 2021).

The recent additions to the Vital set show this is not an LCP-only story. Yahoo! Japan News reduced Cumulative Layout Shift by 0.2 and saw a 15.1% increase in page views per session, a 13.3% increase in session duration, and a 1.72-percentage-point bounce reduction (web.dev Yahoo! Japan News, 2021). Fotocasa, after Interaction to Next Paint became a Core Web Vital in March 2024 (web.dev INP-CWV transition), removed unnecessary rerenders and optimised third-party script execution, growing contact and phone lead conversions by 27% (web.dev Fotocasa, 2024).

A caveat the practitioner literature underuses: Renault's data shows the bounce-rate gain from LCP improvements clusters under the 1.6-second mark (14 percentage points below 1.6s, only 5 above), suggesting performance has thresholds rather than linear returns. Optimise to the threshold, not just below the technical pass mark, if you want the conversion curve to actually move.

Sources