Lighthouse required-N calculator

Use a pilot-estimated run-level standard deviation for the same metric, throttling mode, site class, cache policy, device profile, and hosting environment. This calculator is for planning and review. Empirical bootstrap is still recommended for final median-run claims.

Claim type Run-level SD, sigma Target detectable change, delta

Two-sided alpha Power Candidate runs per condition

Required runs per condition

Candidate-n diagnostics

Approx 95% CI half-width for mean before/after difference	-
Approx 80% power MDE at candidate n	-
Signal-to-noise ratio delta/sigma	-

Run-count sweep

n	CI half-width	MDE