How We Research

Every product fact on this site comes from one place: the manufacturer’s own listing, retrieved through Amazon’s product data API. No spec is typed in from memory, and none is copied from another review site.

What we compare, and why

Models in a guide are set against the same measures, so the table shows real differences rather than each brand’s favorite number.

• Usable capacity, and how it is split. Ten quarts in one cavity and ten quarts across two baskets are different appliances. The guide says which is which.
• Temperature range. The ceiling decides whether you brown food or properly crisp it. The floor decides whether you can dehydrate at all.
• Cooking programs. Counted as the manufacturer counts them, with a note when a “9-in-1” turns out to be seven cooking presets plus preheat and keep-warm.
• Published dimensions and weight. Taken from the listing’s own product information, because counter space is the constraint people forget.
• Basket layout. One cavity or two zones, since that is the single biggest difference in how an air fryer is used.

Where the numbers stop

Several manufacturers do not publish a wattage figure. Rather than fill the gap with a plausible number, the spec table prints not published. The same rule applies anywhere else a value is missing.

What we do not measure

We have not cooked with these machines. There is no lab, no thirty-day trial, no hands-on verdict, and you will not find those words on this site. That also means no score and no star rating. Amazon’s data feed returns no customer ratings to us, so publishing one would mean making it up.

What you get instead is a like-for-like comparison of documented facts, plus our reading of which buyer each model suits. That reading is editorial judgement, and we say so where we make it.

Picking a winner

A “Best Overall” label means one thing: on the measures above, that model suits the widest range of households in the group. It is not the most expensive, not the most reviewed, and it has nothing to do with commission. Where a cheaper model matches the specification that matters, we say that too.

Keeping pages current

Product data is re-fetched from the API rather than edited by hand, so a refreshed page reflects the current listing. When a model is discontinued or replaced we change the recommendation instead of leaving a dead link with a new date on it.