Fieldtested methodology

This page describes the cross-vertical principles. For vertical-specific scoring (AI Agents Power Index composition, weights, sources), see /agents/methodology/.

Core principles

1. Field-tested over lab-tested. When operational evidence contradicts a benchmark, we publish operational evidence. Field Reports flag explicit production context.
2. Sources cited inline. Every quantitative claim links to its source with the date of capture. Frontmatter sources entries are required for reviews and indices.
3. Methodology is versioned. The Power Index methodology has a public version number. Major changes increment the version and trigger a recompute with both versions visible in the archive.
4. Right of reply. Vendors challenged by a review or low ranking can request a response that we publish alongside the original, time-stamped.
5. Affiliate links are disclosed. See disclosure.

What we don't do

• Pay-for-placement, pay-for-scores, pay-for-reviews
• Sponsored content masquerading as editorial
• Programmatic SEO ("X vs Y" pages generated without real analysis)