Taste and culture

The Panda Score: How We Rate Chocolate

By Oliver BennettPublished October 15, 2026Updated October 15, 20265 min read

The Panda Score is one number out of 10, with one decimal, such as 7.7. At least two tasters score a bar blind on five criteria, each as a whole number from 1 to 10, and the Panda Score is the weighted average of their scores. Flavor carries the most weight, and an off-flavor caps it.

Chocolate reviews on this site carry a Panda Score. The five criteria follow the tasting steps in how to taste chocolate: look, snap, smell, melt, flavor and finish. More on tasting is in the taste section.

The five criteria and their weights

Flavor counts for 35% of the score, texture and melt for 20%, and appearance and snap, aroma and finish for 15% each.

Flavor is judged across the whole melt: the first notes, the middle and what’s left at the end, the three stretches the Cacao of Excellence guide asks tasters to notice.

Snap is only compared within one type of chocolate. The research on a strong, clean snap, such as a 2025 study by Stobbs and colleagues, is about dark chocolate, and a white bar isn’t expected to crack like an 80% dark.

Finish is scored on length and cleanliness together. A long finish isn’t automatically good, because a defect can linger as well.

How the score is calculated

The Panda Score is the weighted average of the five criteria, rounded to one decimal at the very end. The order of the arithmetic is fixed:

1. For each criterion, the tasters’ scores are averaged, without rounding.
2. Each average is multiplied by its weight.
3. The five results are added up, and only that total is rounded to one decimal.

So if two tasters give 7 and 6 for aroma, the average of 6.5 goes into the sum as it is, as in the worked example below.

The off-flavor cap

One rule overrides the rest. If a taster finds an off-flavor, such as mold, smoke, or soapy or stale fat, that taster’s Flavor score can be no higher than 4. The cap applies to each taster’s own Flavor score, before the tasters’ scores are averaged.

The off-flavors are the faults at the bottom of the tasting vocabulary: musty, moldy, harsh smoke, cardboard, stale or rancid fat, soapy. They are faults in the bar, so one is enough to pull the Flavor score down.

A worked example

This is an illustration, not a real bar. Two tasters score the same coded sample:

Weighting the averages: 8 × 0.15 = 1.2, 6.5 × 0.15 = 0.975, 8.5 × 0.20 = 1.7, 8 × 0.35 = 2.8 and 7 × 0.15 = 1.05. They add up to 7.725, so the Panda Score is 7.7.

Now suppose taster A had found a musty note. A’s Flavor score could be at most 4, and A’s own result would fall from 7.9 to 6.5, because Flavor carries 35% of the weight: 4 × 0.35 = 1.4 instead of 8 × 0.35 = 2.8. A flawed bar can’t hide behind a nice snap.

How every score is made

The scoring protocol is the same for every review:

• At least two tasters score each bar, and the Panda Score is their average. The tasters are named on the review.
• Tasting is blind. Samples are coded and the brand is hidden. No more than six samples in one session.
• Bars are tasted at room temperature, from their own packaging, and the best-before date is recorded.
• Each bar is judged within its category. Dark, milk, white and ruby chocolate are scored on separate lists, never against each other.
• The score sheet with every criterion score and the notes is kept, and the review shows all five criterion scores alongside the total.

Two tasters is the minimum because people taste the same bar differently. In a 2019 study by Acierno and colleagues that measured the aroma reaching ten people’s noses as they ate ten chocolates, the differences between the people were larger than the differences between the cocoa beans. Sensitivity to grit varies from person to person as well, which how to taste chocolate covers in its melt step.

The six-sample limit follows the 2023 Cacao of Excellence guide, which recommends no more than six chocolate samples in a single session because senses tire and adapt. Coding the samples follows professional practice too: Cacao of Excellence codes its samples, and a University of Reading classroom activity uses random three-digit codes to avoid any bias toward “number 1”.

Why it works this way

Competitions score each product on its own instead of ranking products against each other. The Academy of Chocolate says each product is “not ranked against others, but is scored on a range of attributes and criteria” by several judges, and lists attributes such as aroma, texture, formulation, taste, execution, mouthfeel and aftertaste. Appearance isn’t a separate criterion there. Our “Appearance and snap” criterion is our own choice.

Professional cacao panels go much further. Cacao of Excellence scores core, complementary and off-flavor attributes, each with sub-attributes, on a 0 to 10 intensity scale, and calibrates its tasters with known reference samples before each session. That’s too detailed for a bar review. The Panda Score is a simpler version of the same idea.

How competitions turn those judgments into one number isn’t public. The Academy of Chocolate publishes its five-point scale, from excellent to very poor, but calls the formula that converts it to a total score proprietary. The International Chocolate Awards describes its approach only as statistical analysis of judges’ blind evaluations. So the 1 to 10 scale and the weights above are our own decision. No standard sets them.

We use 10 points rather than five stars because a number reads the same in every language, and one decimal can separate two bars that would share the same number of stars.

Independence

No advertiser or banner sponsor can affect a score, a ranking or a word of a review. Advertising on this site is limited to banners and never appears in the content. If a maker sent us the bar, the review says so clearly. We don’t use affiliate links. When a score changes, say for a new batch or a new recipe, the old score and its date stay on the page.

To taste along at home with the same steps, start with how to taste chocolate.