Field data vs lab data: measuring real-user speed honestly
Short answer: Lab tests are reproducible; field data reflects real devices and networks—use both, trust field data for “how customers feel it”.
NZ owners sometimes optimise until Lighthouse hits 100 while rural visitors on older phones still suffer.
Lab tools
- Great for catching obvious regressions before deploy.
- Controlled environment—can be gamed with unrealistic throttling settings.
Field tools
Chrome UX Report, Search Console Core Web Vitals, and RUM products show distributions—not single scores.
Decision rule
If field metrics are poor at the 75th percentile, prioritise fixes there even if lab looks fine.
Frequently asked questions
Why do scores differ daily?
Traffic mix, device upgrades, and third-party variability all move numbers.
Is one “good” URL enough?
Evaluate templates and page types—home may be fine while product templates fail.