Write down the claim before opening the paper
Start with a complete sentence: what is supposed to change, for whom, after what use of which product? A statement about supporting wellness is harder to test than a statement about a defined symptom score over a stated period. If the claim keeps changing during the conversation, the evidence cannot be matched fairly.
For example, a product might advertise digestive comfort, while a linked paper measures the relative abundance of bacteria in stool. Those outcomes may be related in a research question, but they are not interchangeable. A change in a laboratory measurement is not automatically a change the participant felt, and neither automatically demonstrates treatment of a digestive disease.
NCCIH’s Know the Science resources encourage attention to study design and the difference between anecdotes and evidence. The purpose is not to reject every small study. It is to describe each study at the level it can support, including what a later, larger or differently designed study would still need to answer.
Check what was actually given
Find the product name, formulation, amount and schedule in the methods or supplement information. A study of isolated inulin is not a trial of a powder containing an undisclosed amount of inulin among many ingredients. The same goes for a vitamin, an ashwagandha extract or a named probiotic strain.
A commercial name alone may not establish a formula match across years. Our Green Vibrance review documents a visible version change, while the KOS Apple file separates the current panel from older product naming. Before applying an old paper to a new package, ask the manufacturer to identify the match and any relevant changes.
This is also why a statement that a formula was developed using science is different from a completed human trial of that formula. Ingredient research can inform development. It cannot be relabeled as a direct test of the whole product merely because the ingredients now share a scoop.
Identify the people and the comparison
Look for participant ages, health status, sample size and exclusions. A study in healthy volunteers may not answer a question about people with a diagnosed condition, multiple medicines or a substantially different age range. A page aimed at readers over 50 should not quietly turn a general adult study into dedicated evidence for older adults.
Random assignment and a comparison group help investigators distinguish treatment effects from changes that might occur anyway. A placebo is designed to provide a comparison without the tested active intervention. NCCIH explains why expectations can affect reported outcomes, which makes blinding and the details of the comparison especially relevant to symptom claims.
A trial without a placebo can still provide information, depending on its question and design. It simply cannot answer all the same questions as a well-conducted blinded comparison. Likewise, a customer review is evidence that someone reported an experience, not proof that the powder caused it.
A small published example, read narrowly
A 2024 AG1 randomized, double-blind, placebo-controlled trial described 30 healthy adults and four weeks of supplementation. The full text specifies eligibility ages 18–50, so this is not dedicated evidence for adults over 50. The PubMed record reports microbiome analyses, digestive questionnaires, bowel measures and laboratory safety markers. It is a human trial of the studied AG1 preparation, which is more directly relevant than research on an isolated plant.
That still leaves important limits. The abstract reports a possible digestive quality-of-life improvement with a p value of 0.058; that should not be restated as a conventionally statistically significant benefit. Its short-term laboratory findings also do not establish lifetime safety, safety for every medical condition or the absence of rare adverse effects.
The paper was funded in part by AG1, and the record discloses that two authors were employees of Athletic Greens International. That connection should remain visible. It does not automatically invalidate the work, but it is relevant context when judging independence and the need for replication. We did not establish that every current AG1 variant is identical to the study preparation.
Read results before the conclusion slogan
A paper may measure many outcomes. Notice which ones were planned as primary outcomes, whether the comparison was between groups and whether uncertainty is reported. Improvement from a group’s own starting point does not always mean it improved more than the comparison group.
Percentages also need a denominator and a measurement scale. A large percentage change from a small starting score may have a different practical meaning than the headline suggests. When those details are unavailable in an abstract, say the abstract is the limit of the review rather than reconstructing the missing methods.
Statistical significance and practical importance answer different questions. A result can be statistically convincing yet too small to matter much to daily life; a promising estimate can remain uncertain in a small study. This guide is not a substitute for a full technical appraisal, but these distinctions prevent the most common shopping-page leaps.
Look for the safety window and the funding
Record how long participants were followed, what adverse effects were assessed and who was excluded. An absence of detected problems in a brief, small study is reassuring only within a limited observation. It cannot rule out every interaction or an uncommon harm that the study was too small to detect.
Funding and author interests belong in the evidence note alongside the result. A manufacturer-funded study can still be carefully conducted; an independent study can still have design weaknesses. Judge the methods and disclosures together rather than awarding automatic credibility based on either a brand name or an institutional logo.
Our ashwagandha guide shows why preparation and duration matter for an ingredient. The detox-claims guide asks a complementary question: whether the advertised outcome is defined well enough for a study to test at all.
Bring the conclusion back to the exact purchase
A useful evidence note names the study, tested formula, population, comparison, duration, measured outcome and unresolved limitation. It should be short enough to compare with the seller’s sentence. If they do not match, ask for the missing evidence instead of widening the paper’s conclusion yourself.
The sponsored CoreAge Green Scene review follows the same distinction between label description and proven benefit. First placement, an organic seal, a large ingredient list and a research link each communicate different information. Keeping them separate makes a review more useful than a score that hides how it was reached.