Start with a repeatable measurement

Choose the homepage, a service page and a contact journey. Record the URL, device profile, connection settings and test date before changing anything. Repeat the same test afterwards. A Lighthouse lab result can reveal bottlenecks; it is not a substitute for real-user data. Core Web Vitals guidance uses LCP at or below 2.5 seconds, INP at or below 200 milliseconds and CLS at or below 0.1 at the 75th percentile.

Send an image that fits the screen

A phone should not download a desktop-sized image when a smaller version is sufficient. Generate several widths, use srcset and sizes, and keep explicit dimensions so layout space is reserved. For example, a 390px display at two device pixels per CSS pixel may need an image around 780px wide. Check the browser’s selected source rather than assuming the smallest file is always used. Prioritise the main visible image and lazy-load images further down the page.

Give motion a job and an exit

Use motion to connect states or guide attention. Avoid hiding essential content until a scroll animation runs. Respect the device’s reduced-motion preference and provide a pause control for continuous movement. Check that disabling animation restores normal document flow, especially for pinned horizontal sections. Keyboard users should still reach every link and control without navigating a moving target.

Check the whole journey after a change

Resize the viewport, navigate using only the keyboard and submit an invalid form to inspect its errors. Test the page without JavaScript for essential text and contact options. Watch for layout movement when fonts and images load. Review transferred bytes and long tasks as well as the headline score. If field data is unavailable because traffic is low, record that limitation and keep using reproducible lab checks.

Your practical checklist

  • Record the test URL, device and connection
  • Provide responsive image sources and dimensions
  • Keep important content visible without animation
  • Verify reduced motion, pause and keyboard focus
  • Compare the same lab setup before and after changes
  • Review field data when sufficient traffic is available

Services that can help

Read next

Working through a similar question?

We’re happy to help you find a practical starting point.

Talk to AMGK