Classic Latte

Screening model notes

How Nutri-Score, NOVA, and list length read this record

Within the four-tier shelf-label framework this site publishes, the composite for this product falls in the highest band. That band is defined by a threshold on the derived score alone: it is where labels land once components carrying caution-or-worse registry ratings contribute little or nothing to the deduction. The band is a summary of what the screened registries return for the components that were parsed, so it inherits their coverage gaps — a component absent from both registries neither raises nor lowers the figure. The threshold and the weighting behind it are set out on the methodology page.

Nutri-Score settled on its middle grade here, the band the algorithm produces when the penalty and credit sides of its nutrient ledger come out close to balanced. A mid-scale grade is a statement about the panel as a whole rather than about any single nutrient, and products arrive at it by very different routes: a moderate profile across every term, or high marks on some offset by heavy penalties on others. The grade is computed separately from the ingredient screen below and the two are not comparable measures.

The declared list parsed into a small number of components. On a short list the registry screen covers close to everything the packaging declares, and any single matched component accounts for a correspondingly large share of the composite. Short lists also leave less room for parsing error, since there are fewer separators and qualifiers for the extraction to resolve.