How we review

Three inputs, in a strict order.

1. Measured accuracy, with published methodology

Every accuracy figure we cite has a published methodology behind it: the meal set, the reference method, the statistical procedure. The anchor is the 180-meal weighed-food validation of six commercial apps, its per-complexity breakdowns, and the independent reproduction of the top result on a second 231-meal set. Figures without methodology — vendor marketing numbers — are treated as unevaluated, and we say so.

2. What the free tier actually gates

A tracker's paywall placement is an editorial decision by its maker. We map what is free and what is paid on every plan, because a free tier that gates precision (a barcode scanner) is a different product decision than one that gates convenience (unlimited photo scans).

3. Adherence reality

The best app abandoned in week three has an effective error of 100%. Logging speed, interface friction and the workflow's survival past week two are part of the review, not a footnote.

What we don't do

• We accept no payment, affiliate commission, or free units from app makers.
• We don't reprint vendor accuracy claims without their methodology.
• We don't rank by feature count; we rank by whether the features are measured.

Corrections

When new weighed-food data publishes, we update the ranking and date the change. When we're wrong, the correction appears on the page.