How to pass Core Web Vitals

Executive summary

Core Web Vitals measure loading (LCP), interactivity (INP) and visual stability (CLS) from real users. Optimise against field data, not lab scores, and fix the metric that is failing for the most templates first. Most wins come from the largest image, the heaviest script and unsized layout elements.

What this helps you decide

Which of the three vitals to prioritise, and which specific fix will move real-user data across the threshold.

Business problem

Slow, unstable pages frustrate users and sit below the passing threshold for Core Web Vitals, costing conversions and a modest but real ranking signal. Teams often optimise blind, chasing lab scores that never move the real-world field data.

Step-by-step process

1. 1
   Read the field data first

   Use the Search Console Core Web Vitals report and real-user data, not one lab test, to see which metric fails on which page groups.

2. 2
   Fix the largest contentful element

   For LCP, identify the largest above-the-fold element, then serve it efficiently: compress and correctly size the hero image and preload it if needed.

3. 3
   Cut and defer JavaScript

   For INP, reduce main-thread work by removing unused scripts, splitting long tasks and deferring non-critical third parties.

4. 4
   Reserve space for dynamic content

   For CLS, set explicit dimensions on images, ads and embeds so nothing shifts as the page loads.

5. 5
   Optimise delivery

   Enable caching, compression and a content delivery network so bytes arrive quickly to real users everywhere.

6. 6
   Validate on real devices

   Test on a mid-range mobile device on a throttled connection, because that is closer to how the field data is gathered than a fast desktop.

7. 7
   Watch the 28-day trend

   Field data updates on a rolling window, so confirm the passing rate climbs over several weeks rather than expecting an instant flip.

Worked example

Checklist

• Diagnose from field data, not a single lab run
• Compress, size and preload the largest contentful element
• Remove or defer non-critical JavaScript to improve INP
• Set explicit dimensions on images, ads and embeds
• Enable caching, compression and a CDN
• Confirm the passing rate improves over the 28-day window

Common mistakes

• Chasing a perfect lab score while real-user field data stays red
• Optimising one page when the failing metric is a template-wide problem
• Adding a heavy third-party script that quietly wrecks INP after the fixes ship

30-minute experiment

KPIs to track

• Percentage of URLs passing all three vitals in field data
• LCP and INP at the 75th percentile
• Conversion rate on improved templates

FAQs