The Ingredient Brief

Evidence

Why Evidence for an Ingredient Is Not Evidence for a Formula

Published July 12, 2026 · The Ingredient Brief

This is the most expensive inference error in the supplement aisle, and almost nobody has to lie to make it happen. A real study is run on 500 mg of an isolated compound. A bottle contains an undisclosed fraction of that compound alongside eight other things. The citation is accurate. The transfer of its authority to the product is not.

The move, in slow motion

Marketing copy rarely claims that the product was studied. It says the product contains an ingredient that was studied — which is true — and then places the two sentences close enough together that your mind does the rest. The upgrade from “evidence about a compound” to “evidence about this bottle” happens silently, in the reader's head, where nothing can be fact-checked.

The reason this works is that the upgrade sounds like arithmetic. If the compound does something, and the compound is in there, then the bottle does something. The trouble is that every step of that sentence hides a variable that the research fixed and the product changed.

Four things that break the transfer

1. Dose

Research is dose-attached. A trial does not find that a compound “works”; it finds that a particular amount produced a measurable difference against a control. Cut that amount to a fifth and you no longer have a weaker version of the finding — you have an untested condition. The honest description of a below-study dose is not “less effective.” It is unknown, which is a different and less comfortable word.

This is also where the label often stops cooperating. Under US rules, a proprietary blend may declare only the total weight of the blend, listing its ingredients in descending order by weight without disclosing any individual amount. When the dose is the whole question and the panel withholds the dose, the comparison you would need to make is not merely inconvenient — it is unavailable.

2. Chemical form

“Magnesium” is not an ingredient. It is an element that arrives inside a salt, and the salt decides how much of it ever reaches your bloodstream. In a randomised, double-blind trial of 300 mg daily over 60 days, magnesium citrate and an amino-acid chelate were absorbed measurably better than magnesium oxide, with the oxide arm showing no meaningful separation from placebo on the absorption markers used. Same element. Same milligrams on the panel. Different outcome in the body.

So when a study is run on one form and a label lists another — or lists the element with no form named at all — the number on the panel and the number in the study are not the same quantity. They only look alike.

3. The matrix, and what sits next to it

Ingredients do not act in isolation once they are in the same capsule. Sometimes a co-ingredient dramatically raises absorption: in a classic pharmacokinetic study, 20 mg of piperine taken with 2 g of curcumin raised curcumin's bioavailability in human volunteers by an order of magnitude relative to curcumin alone. Sometimes the interaction runs the other way, with compounds competing for the same absorption pathways.

Either way, the point stands: an ingredient's behaviour inside a specific formula is an empirical question about that formula. It cannot be read off a study of the ingredient by itself. The Federal Trade Commission has litigated exactly this, accepting expert testimony that trials on a supplement should use “the same dosage and formulation rather than … individual ingredients because there may be interactions between the ingredients that affect their physiological actions.”

4. The finished product usually was not tested

The quiet fact underneath all of the above: testing a finished multi-ingredient formula in humans is expensive, slow, and — for most products — was simply never done. What exists instead is a literature about the components, assembled into a citation list that reads like evidence about the whole. A bibliography is not a trial.

The regulator's own checklist is the tool

You do not have to invent a standard here. The FTC's Health Products Compliance Guidance tells advertisers what questions their own evidence must survive before they may make a claim. Those questions work just as well pointed in the other direction — from the buyer at the shelf back at the label. The guidance asks advertisers, in its own words:

  • “How do the dosage and formulation of the advertised product compare to the product used in the study?”
  • “Is the ingredient or combination of ingredients in the advertised product the same as what was used in the study?”
  • “Is the advertised product administered in the same manner as the product in the study?”
  • “How well do the outcomes tested in the study relate to the specific benefits advertised?”
  • “Does the study population reflect the characteristics of the population targeted by the ad?”

Collapse those into one sentence and you have the only question this article is really about: was the study run on this — this dose, this form, this combination, in people like me?

What to do when the answer is no

It will be no, most of the time. That is not a reason to throw the aisle out. It is a reason to stop treating a citation as a verdict and start treating it as what it is: a claim about an ingredient that the product has not yet earned. Three moves follow.

Downgrade the claim from demonstrated to plausible. Say it in those words. A product whose evidence is ingredient-level is not disproven — it is unproven at the level you are buying. Holding that distinction is the entire skill. It is what keeps you from dismissing a decent formula and from trusting a hollow one.

Then check what is still checkable. The formula may be untested, but plenty of the product remains inspectable, and this is where you recover most of your leverage. Is every ingredient's amount disclosed, or hidden in a blend? Is the chemical form named, or only the element? Where a studied dose exists, does the label dose reach it — or is it arithmetically impossible given the blend's total weight? Is there third-party testing confirming the panel is accurate? A formula that answers all four has not proven it works, but it has proven it is willing to be measured, and that is a real and rare signal.

Finally, notice what the seller is competing on. A formulator who has run a trial on the finished product will tell you so, loudly and specifically — product name, dose, duration, comparison group. Vagueness about the study is almost never an accident of writing. When a page cites nine studies and none of them are on the thing in the bottle, that pattern is itself the finding.

The honest bottom line

None of this proves that multi-ingredient formulas do not work. Some of them plausibly do, and the absence of a finished-product trial is a gap in the evidence, not proof of absence. But somebody has to pay for that gap, and in this market it is not the manufacturer — the citations cost nothing to reprint. It is the buyer, in money and in months.

You close the gap by asking one question at the shelf, out loud if it helps: was this study run on this? When the answer is yes, the evidence travels. When it is no — and it usually is — you have not lost anything. You have simply stopped paying full price for a claim the label never actually made.

The short version

  • An ingredient study is evidence about a dose, a chemical form, and a preparation — change any of the three and the evidence stops applying.
  • Interactions are real: the FTC itself has accepted that ingredients in a formula can alter one another's physiological action, so component studies cannot stand in for the formula.
  • Run the check: “was the study run on THIS — this dose, this form, this combination, in people like me?” Almost always the answer is no.
  • When it is no, downgrade demonstrated to plausible, then judge the product on what remains checkable: disclosed doses, named forms, arithmetic that fits, third-party testing.

Sources