Written from the trenches of Auckland real estate by Amit Sharma — Bayleys agent, 10+ years marketing experience.
Almonds contain 579 calories per 100 grams. A cup of whole almonds contains 828. Both numbers are true, both come from the same USDA record, and only one of them describes anything a person has ever eaten. That gap, from 579 to 828, is the entire problem with nutrition labels in a single food, and once you notice it you cannot stop noticing it.
I built a reference site called FuelFacts partly to stop making this mistake myself. It publishes 1,799 foods with the full nutrient panel, 15 nutrients each, drawn from USDA FoodData Central, specifically the SR Legacy release, which is public-domain United States government material. The one editorial decision I made was to show every food two ways at once: per 100 grams, and per the actual household serving USDA records for it. Almonds, for instance, come with 1 cup whole at 143.0 g, 1 cup sliced at 92.0 g, 1 cup slivered at 108.0 g and 1 cup ground at 95.0 g. Four cups, four different weights, four different calorie counts, all of the same nut.
Per 100 grams exists for a good reason, and it is worth defending before criticising it. It is the only way to compare two foods honestly, because it holds the denominator constant. Without it, a manufacturer can make almost any product look favourable simply by choosing a flattering portion. The site puts this bluntly: serving sizes are chosen by marketers. A snack bar declared as two servings, a bottle declared as 2.5 servings, a cereal measured in a portion nobody has ever poured. The per-100g column is the defence against all of that, and regulators mandate it for exactly this reason.
The problem is that per 100 grams solves the comparison problem and creates a perception problem. Nobody eats 100 grams of anything by intention. You eat a handful, a cup, a slice, a fillet, a bowl. So the number that is technically the most comparable is also the number least connected to your day, and your brain still quietly treats it as a portion because it is the number printed largest.
This distortion runs in both directions, which is what makes it genuinely tricky rather than just annoying. Foods eaten in small amounts look alarming per 100 grams. Spices and herbs are a clean example: it is one of the 23 categories on the site, with 15 items in it, and per 100 grams several of them are extraordinarily dense in minerals. In practice you use a gram. Reading that column as though you were going to eat a small bowl of dried herbs produces a completely false picture of both the benefit and the cost.
Foods eaten in large amounts do the reverse, and this is the direction that actually costs people. Anything with high water content, so drinks, soups, sauces and many prepared dishes, has a modest per-100g figure by construction, because a large share of that 100 grams is water. Then you consume three or four times that weight in one sitting without registering it as unusual. Nothing on the label was false. The multiplication just never happened.
Almonds sit in a third and sneakier category: calorie-dense foods eaten by volume rather than weight. Nobody weighs nuts. You take a handful, or you pour a cup while cooking. At 579 kcal per 100 g, a cup of whole almonds is 143 grams and 828 calories, with 30.2 g of protein and 71.4 g of fat. The per-100g number is not wrong. It is just about seventy percent of what actually left the jar. And because the food is genuinely healthy, there is no warning signal to make you check.
So what should you actually use? Both columns, for different jobs, and never one alone. Use per 100 grams when you are choosing between two options: which yoghurt, which bread, which cut of meat. That is a comparison question and it needs a constant denominator. Use the per-serving figure, in the household measure you genuinely use, when you are estimating what you have eaten today. That is an accounting question and it needs your real portion.
The habit worth building is small: after reading a per-100g figure, do one multiplication before forming an opinion. If your portion is 150 grams, the real number is one and a half times what you just read. If it is 40 grams, it is under half. It takes two seconds and it changes conclusions more often than you would expect. Most people are not bad at nutrition. They are just skipping one multiplication.
It is also worth knowing how much precision any of these figures deserve. Reference values like these are averages for a food type, and real food varies by brand, ripeness and preparation, typically in the region of ten to twenty percent from the reference. That is not a reason to distrust the data. It is a reason to treat the third significant figure as decoration and the order of magnitude as the actual message. Whether almonds are 579 or 620 per 100 grams matters far less than whether you had 30 grams of them or 143.
Rankings are where the choice of denominator does the most damage, because a league table with the wrong denominator produces confidently wrong advice. The site has ten of them across all foods, highest protein, lowest calorie, highest fibre, highest potassium, highest iron, highest calcium, highest vitamin C, lowest carb and lowest sodium among them, plus protein, calorie and fibre rankings inside 13 individual food categories. Every one of them is computed per 100 grams, and it says so on the page. Read as a comparison, they are useful. Read as a shopping list of things to eat by the bowl, they will mislead you, and the fix is the same multiplication.
If you want to see a food both ways, the database is at nutrition data for 1,799 foods shown per 100 g and per real serving, with the USDA record ID on each entry so you can trace any number back to the source rather than taking my word for it. It is free, there is no account, and it sits alongside the other small reference tools on my tools page.
The short version is that neither number is a lie and neither number is enough. Per 100 grams tells you what a food is. Per serving tells you what you did. You need both to say anything useful, and the ten seconds it takes to hold them side by side is the highest-return habit in this whole subject.