Methodology

Repeatability before spectacle.

Every plugin is installed on the same controlled lab, using default settings and the same measured URLs.

Plugins are activated as shipped. Required onboarding uses default or pre-selected choices only. No manual performance tuning is applied.

Each plugin, URL, and tool combination is measured five times. Public summaries are derived from raw runs using medians and visible ranges.

Individual test runs are authoritative. Summary rows are generated or reviewed aggregates and never replace raw measurements.

The model supports corporate and ecommerce scenarios from the beginning. Phase one publishes only the corporate scenario.

No final formula is hardcoded. Any future experimental score will be secondary metadata and will not hide raw metrics.

The site is visually checked after plugin activation so faster metrics do not conceal a broken page.