Evidence policy
Every claim must show where it came from.
A clear evidence label is more useful than a confident score. We separate public information, controlled measurements, and field observations so buyers and agents can judge each conclusion.
Version 1.0 · Last reviewed July 21, 2026
Evidence levels
1. Public-source profile
Facts are taken from named manufacturer pages, manuals, or other attributed public sources. We normalize terminology and identify contradictions. We do not imply that the product was handled or measured.
2. Bench tested
A result can use this label only when the sample identifier, sample source, procedure, equipment, conditions, and observation date are recorded. A single sample cannot prove production-wide quality.
3. Field observed
Workflow, ergonomics, and durability observations are tied to a defined job, operator context, and period of use. Field observations stay separate from controlled measurement claims.
Product-page requirements
- Evidence level and hands-on status.
- Exact model or variant, when confirmed.
- Source list and last-reviewed date.
- Sample source and commercial relationship.
- Known limits, contradictions, and unanswered questions.
- Who should not choose the product, once evidence supports that conclusion.
Scores and rankings
We do not publish a numeric score until the relevant protocol and evidence are available. Category winners must be tied to a use case. Payment cannot buy a score, ranking position, or the removal of a supported criticism.
Supplied products and affiliate links
If a manufacturer, distributor, or other party supplies a sample, loan, travel, data, or payment, that relationship will be stated on the affected page. Future affiliate links will be marked close to the link. Editorial conclusions remain ours.
Corrections
Readers and manufacturers may submit a correction with a primary source through the research request form. Material corrections will change the page and its reviewed date. We do not silently turn unverified claims into facts.