Source hierarchy
We prioritize exact Amazon listings for product identity and current affiliate destinations, then manufacturer pages, manuals and recognized governing or service sources for technical context. We record a verification date and never borrow a specification from a similar model.
Comparison weights
Our standard framework assigns 25% to room fit, 20% to airflow evidence, 20% to construction, 10% to scoring and useful features, 10% to assembly and ownership burden, 10% to audience fit and 5% to value for the use case. Scores appear only when the underlying fields are documented.
Missing data
Unknown is not zero and does not mean a feature is absent. When the source does not publish a field, we label it not published and ask the shopper to verify it.
Best-for language
We name a best option only for a defined scenario. We do not claim a universal winner, convert voltage into airflow, invent test results or rank products according to commission.
Updates
Marketplace listings change. We show the research date and use the live retailer page for price, seller, availability, shipping and returns.
Last reviewed: September 2, 2026.