How fast should my website actually load?

Core Web Vitals are three field measurements taken from real Chrome users on your site, not from a test rig. Google publishes the thresholds, and they have been stable long enough to build against.

Field data is what counts for ranking. A perfect lab score with poor field data means your test conditions are kinder than the conditions your visitors actually face.

Good is a passing grade, and it is graded at the 75th percentile of your traffic. Hitting it means three quarters of your visitors had an acceptable experience. It says nothing about the quarter who did not.

Build to the target rather than the threshold and the margin absorbs the bad days: the visitor on a weak signal, the older phone, the afternoon your image host is slow.

In practice it is rarely the hosting, which is where most owners look first. It is almost always weight the page did not need to carry.

Uncompressed hero images at desktop dimensions served to phones. A page builder shipping its entire component library to render four sections. Six tracking scripts, three of which nobody has looked at since installation. A font loading in four weights when the design uses two.

Run the URL through PageSpeed Insights. Read the mobile tab. If a field-data panel appears at the top, that is real visitor data and it outranks everything below it. If it does not appear, your traffic is too low for Google to have collected a sample, and the lab score below is your best available estimate.

Then open the page on your own phone, off wifi, and watch what happens in the first two seconds. The stopwatch is honest in a way that dashboards sometimes are not.