A score is not the user experience
Core Web Vitals provide a shared vocabulary for loading, responsiveness, and visual stability. They are useful, but a green dashboard does not prove that users can complete their work comfortably.
Start from real tasks
I connect performance data to journeys such as opening an article, filtering a list, or completing a form. A page can have a good aggregate score while one important interaction remains slow on mid-range mobile hardware.
Field data also needs segmentation. Device capability, network quality, geography, authentication state, and cache status can hide very different experiences behind one percentile.
What I monitor together
- Core Web Vitals from real users
- route and API latency
- JavaScript errors and failed requests
- task completion or abandonment
- device and network segments
The goal is not to maximize a laboratory score. It is to find delays that interrupt user intent, verify a fix with field data, and prevent regressions with a small set of meaningful budgets.