Lab vs Field Data

Lab data comes from a controlled test run against a page, while field data is collected from the browsers of the real visitors who opened that page.

A lab measurement fixes the device, the network speed and the location, so the result is reproducible, can be run before release, and comes with a resource waterfall that names the file at fault. A field measurement fixes nothing: the output is a distribution across the real population of visitors, and that distribution, read at the 75th percentile, is what Google uses to assess Core Web Vitals. Some metrics exist on only one side. Total Blocking Time is meaningful only in a lab because it assumes nobody is interacting, while INP cannot be produced without real interaction. The common mistake is expecting the two numbers to agree; they answer different questions, since the lab explains why a page is slow and the field says for whom it is slow.