Core Web Vitals: What They Measure and How to Fix Them

The three metrics

• LCP (Largest Contentful Paint) — how long until the main content appears. Target under 2.5 seconds.
• INP (Interaction to Next Paint) — how quickly the page responds to taps and clicks. Target under 200 ms.
• CLS (Cumulative Layout Shift) — how much the layout jumps while loading. Target under 0.1.

Field data from real visitors is what counts. Lab tools are for diagnosis; the 75th percentile of real users is the score.

Fixing LCP

• Serve the hero image in a modern format, correctly sized, and never lazy-load it.
• Preload the LCP image and the primary web font.
• Enable full-page caching so HTML arrives immediately.
• Remove render-blocking CSS and inline only the critical styles.
• Cut server response time — this is where hosting quality shows up directly.

Fixing INP

• Reduce total JavaScript; audit what each plugin or tag loads.
• Break long tasks and defer non-essential scripts until after interaction.
• Load third-party tags — chat widgets, heatmaps, ad scripts — asynchronously or on demand.
• Avoid heavy work in scroll and input handlers.
• Replace bloated sliders and animation libraries with CSS where possible.

Fixing CLS

• Set explicit width and height (or aspect-ratio) on all images, videos and embeds.
• Reserve space for ads, banners and cookie notices instead of injecting them above content.
• Preload fonts and use font-display: swap with a matched fallback metric.
• Never insert content above existing content after load.

A repeatable workflow

• 1 Pull field data for your key templates and identify the worst metric.
• 2 Reproduce it in a lab test on a throttled mobile connection.
• 3 Change one thing, measure, and keep or revert.
• 4 Re-check field data after 28 days, since real-user metrics lag deployments.

Frequently asked questions

Do Core Web Vitals affect rankings?

They are a real but modest signal. Their bigger effect is on conversion — faster pages measurably sell more.

Why is my lab score perfect but field data poor?

Lab tests use one device and connection. Field data includes older phones and slower networks, which is what your customers actually use.