Almost every category has one specification that quietly decides whether the thing works. It is almost never the number printed largest on the box — and once you know where to look instead, you stop buying the wrong version of the right product.
Three examples, all of them things I own and use. Every figure below comes from the National Institutes of Health’s own fact sheets or from a published study, not from a manufacturer.
1. The supplement bottle counts something different from what you think
Magnesium is the one everybody gets backwards, including me. The common advice is that the big number on the front overstates what you get, because it is the weight of the whole compound. For a US label that is wrong. NIH states plainly that the Supplement Facts panel declares the amount of elemental magnesium, not the weight of the magnesium-containing compound. The number is honest.
What actually varies is the form. NIH reports that small studies have found magnesium in the aspartate, citrate, lactate and chloride forms is absorbed more completely than oxide and sulfate. Note the hedge — “small studies” — because that is how the source words it. And note what is missing: glycinate, which is the form I take, is not one NIH addresses either way. So I am not going to tell you it is better. I will tell you it is the one in my cupboard and that the evidence I can point to does not cover it.
Fish oil is where the big number really does mislead. The front of the bottle sells you oil. The thing you are buying it for is EPA and DHA. NIH describes a typical 1,000 mg fish oil supplement as containing 180 mg EPA and 120 mg DHA, and notes that doses vary widely between products. Those two lines are the ones to read, and they are never the ones in large type.
Creatine has a similar gap, in the reader’s favor for once. Monohydrate is 88% creatine by weight, and is the most widely used and studied form. NIH notes that the other, usually more expensive forms have not been proven superior to it for raising muscle creatine, digestibility, product stability or safety. Not proven inferior — simply not proven better. Which means the cheapest one on the shelf is the one with the evidence behind it.
And vitamin D is just two scales for one thing. 1 mcg equals 40 IU. A bottle saying 25 mcg and a bottle saying 1,000 IU are the same amount. People buy twice because the units changed.
The case where the label is not the problem — it is whether the label is true
A 2017 study tested 31 melatonin supplements bought from ordinary grocery stores and pharmacies. For most of them the amount of melatonin did not match what was printed on the label, and 26% contained serotonin — a hormone NCCIH notes can have harmful effects even at relatively low levels.
That is the argument for buying the brands that pay for third-party testing, and it is most of what decides the names on my own shelf. It is not a claim that any particular product is contaminated. It is a claim about a category where nobody checks the arithmetic for you.
2. A smart scale is accurate about people and vague about you
It sends a small current up through your feet and infers fat from the resistance. Measured across a group, the method holds up: in a study of 106 adults, foot-to-foot consumer scales averaged within about 1.5 percentage points of a DXA scan.
Measured on one person, the same study put the 95% limits of agreement for the weakest foot-to-foot device at 14.5 points below DXA to 8.6 points above it. The average was good. Your individual reading could be a long way out, and you would have no way of knowing which.
So the number is not the product. The trend is. Across fifteen devices tested against a laboratory four-compartment model, the error in tracking change over time was roughly half the error in any single reading. Even then, the same researchers found about half the devices could not track change dependably and wrote that consumers should be wary of assuming they can.
Which makes the conditions the real instrument. Same time of day, same state, every time. In one four-day study an equilibrated gain of 1.87 litres of body water showed up on the scale as 1.17 kg of fat lost and 1.21 kg of lean gained. None of that happened. It was water.
3. A food log is only as good as the database underneath it
And on most of these apps, that database was filled in by other users.
Dietitians entered 240 two-day food records from 120 adults into five popular apps and compared the results against a clinical reference method. Counting agreement as landing within 5% of the reference, MyFitnessPal matched on energy for 24 of the 120 people. Loosening the threshold to 10% still left every app in the range of roughly half to two-thirds. A separate validation had to discard 2.8% of entries outright as implausible before the comparison could even be run — and once cleaned, its energy figures tracked the reference closely.
So the shape of the finding is this: calories and the big three macros come out usable. Micronutrients and sodium are where these apps fall apart — one study found significant underestimation of calcium, iron and vitamin C. Log to see the pattern in how you actually eat. Do not log to audit a number to the gram.
The rule underneath all three
Read the thing in your hand. Not the headline on the box, not the review, not the number the marketing chose. The elemental figure on the panel, the EPA line rather than the oil weight, the trend rather than today’s reading, the macro rather than the micronutrient.
It is a slower way to shop and it is the only one that survives contact with a product you are going to use for years.