What Core Web Vitals measure
Core Web Vitals focus on three parts of a page experience. Largest Contentful Paint measures when the main content becomes visible. Interaction to Next Paint measures how quickly the page responds to clicks, taps, and keyboard input. Cumulative Layout Shift measures unexpected visual movement.
Google's current good boundaries are 2.5 seconds for LCP, 200 milliseconds for INP, and 0.1 for CLS. Field assessment uses the 75th percentile so the target reflects what most visits experience, not just a fast median.
Field data and lab data answer different questions
Field metrics come from the Chrome User Experience Report and summarize eligible real visits over time. They show whether actual users are receiving a good experience, but require enough public traffic to protect privacy and produce a representative sample.
Lighthouse lab data is generated from one controlled test. It is repeatable enough for diagnostics and can surface render-blocking work, slow server responses, or large resources. It cannot produce a realistic INP without representative user interactions, so field data remains the stronger source for responsiveness.
How to improve a failing result
For LCP, start with server response time, the priority of the main image or text resource, and render-blocking CSS or JavaScript. For INP, find long main-thread tasks and break work into smaller units so the browser can paint promptly after input. For CLS, reserve space for images, embeds, ads, and late-loading interface elements.
Test both mobile and desktop after meaningful releases, but avoid optimizing a single Lighthouse score in isolation. The durable goal is a consistently good experience across real devices, networks, routes, and user journeys.