Web performance checklist
A Core Web Vitals-grounded performance checklist — measurement, loading, images, JavaScript, and fonts — for auditing a real page, not a lab score.
Last updated
Your ticks are saved on this device only — never uploaded, never synced.
Measure first
0/4Loading
0/5Images & media
0/5JavaScript
0/5Fonts & CSS
0/4Questions
Why measure before changing anything?
Because performance intuition is unreliable even for people who wrote the code, and without a baseline you cannot show a change helped. Optimising the wrong thing is invisible work that produces nothing.
Lab data or field data?
Both, for different questions. Lab tools are reproducible and let you compare changes; field data is what real users experience on real devices and networks, and the two frequently disagree.
What usually gives the biggest win?
Images and JavaScript, in that order, on most sites. Correctly sized modern-format images and removing unused script routinely beat any amount of micro-optimisation elsewhere.
Does a perfect score mean a fast site?
No. A synthetic score is a proxy measured on one configuration, and a site can score well while feeling slow on a mid-range phone. Field measurements are the ones that describe your users.
How do I stop it regressing?
Budget it in CI, so a change that exceeds a size or timing threshold fails the build. Performance won by a project is lost gradually by ordinary work unless something enforces it.