🧪 Methodology

Weighted rubric

• Quiz depth: 25%
• Author and learner UX: 20%
• Logic, results, and reporting: 20%
• Maintenance and trust: 15%
• Integrations and data ownership: 10%
• Value: 10%

Maintenance uses captured WordPress.org recency, active installs, rating, rating count, support resolution, and tested-version compatibility. Other categories combine official documentation with the live free-tier bench; live behavior supersedes documentation where they conflict.

Evidence tiers

• Live roster: fourteen pinned WordPress.org packages installed on isolated multisite subsites and run in Chromium.
• Live controls: three pinned WordPress.org packages installed and activated on their own subsites whose free build authors no question in WordPress; only the observed boundary is credited and no score is issued.
• Documentation tier: commercial, add-on, or hosted products evaluated only from official sources; no capability is presented as live proof.
• Candidate discovery: complete WordPress.org result pagination for quiz, survey, exam, assessment, trivia, flashcard, poll, and test, then first-class quiz-authoring filtering.

Free-tier and safety boundary

No paid or nulled code was installed. No vendor package outside WordPress.org was used except an official H5P.org content library package imported through H5P’s own library interface. Paid capabilities are labelled pro-gated; unavailable capabilities are unsupported.

Completion rule

A demo passes only when the native interface renders the benchmark, a browser submits an attempt, and a native result is observed. Forminator’s failed persistence, Tutor’s blocked learner transition, and the login walls in Sensei LMS and Academy LMS are reported as failures rather than repaired with fake markup. From 2026-08-19 the completion check runs in a clean logged-out browser context, so an authenticated attempt alone never counts as a pass.

Full method, benchmark questions, measurement contract, and evidence rules.