Supplement Claim Substantiation: What Actually Counts as Evidence

Your formulator says the sleep blend works. Your agency wants "clinically studied" on the label. The Amazon listing goes live in six weeks.

Nobody in that chain has answered the two questions supplement claim substantiation turns on: which kind of claim the product may make, and what evidence must exist before you advertise it.

Below is what the Food and Drug Administration (FDA) and the Federal Trade Commission (FTC) each expect, where substantiation usually breaks, and how to pick the study your claim needs.

The short answer

• Two agencies, two questions. Both expect claims to be supported by evidence. FDA primarily oversees labeling, while the FTC primarily oversees advertising.
• The FTC bar is high. Health benefit claims generally need randomized, controlled human trials.
• Lab, animal, and informal survey data aren't enough on their own. Validated questionnaires, however, can contribute meaningful evidence when used appropriately in well-designed human trials.
• Calling a study a "pilot" doesn't fix an unsupported health claim. The claim still needs adequate evidence, and any limitations need to be clearly explained.
• Fix it before enrollment. Set the endpoint, sample size, and analysis plan in advance, on the product or ingredient you sell.

Two regulators ask two different questions

Brands tend to treat substantiation as one problem. There are two, enforced by different agencies against different things.

FDA decides what kind of claim the label can make

The first question is which type of claim you are making.

FDA explains the distinction on its structure/function claims page. A supplement carrying a structure/function claim then has four obligations:

1. Notify FDA within 30 days. Under 21 CFR 101.93, the maker must notify FDA no later than 30 days after it starts marketing the product with the claim.
2. Carry the disclaimer. The label states that FDA has not evaluated the claim, and that the product is not meant for diagnosing, treating, curing, or preventing disease.
3. Hold the evidence yourself. FDA does not review your evidence before the claim appears, so the manufacturer must hold substantiation that the claim is truthful and not misleading.
4. Expect the FTC's yardstick. FDA's 2009 substantiation guidance says the agency intends to judge that evidence by a standard consistent with the FTC's.

The FTC decides what evidence sits behind the advertising

Every health claim you run in an ad, on a product listing, on a landing page, or in a social post needs substantiation before it runs. The FTC's Health Products Compliance Guidance, issued in December 2022 to replace its 1998 supplement advertising guide, calls the standard "competent and reliable scientific evidence."

In plain terms, the research has to be designed and assessed objectively by people with expertise in the relevant field, using methods that experts in that field accept as producing accurate results. The evidence must be assessed alongside the full body of relevant, reliable research. A favorable study cannot be considered in isolation from credible conflicting findings.

Which category a specific claim falls into, and what it needs behind it, are calls to make with qualified regulatory counsel. What this guide describes is how the rules are written, as general information rather than legal advice.

What the FTC expects from a study

The guidance gets specific at this point, and most brands discover here that their evidence falls short. Randomized, controlled human trials (RCTs) sit at the top of the FTC's evidence hierarchy for health benefit claims, and the guidance then lists the design principles it uses to judge any study.

• A control group. Results in the product group are compared against a control group, ideally on a placebo or sham, so the product's effect can be separated from expectation, practice, and the passage of time.
• Randomization. Participants are assigned to groups by chance to reduce selection bias and help balance factors such as age, diet, and health.
• Double blinding. Neither the participants nor the researchers know who is in which group, wherever the design allows it.
• A statistically significant difference between groups. The finding used to support the health-benefit claim has to show a statistically significant difference between product and control. Not every measured outcome must reach significance, but a study that measures many outcomes should report all of them and adjust for the extra chances of a false positive.
• A clinically meaningful effect. A statistically significant change can still be too small to make any real difference to the people buying the product, and the FTC expects a claim to clear both bars.

Read the control-group principle closely. A before-and-after change inside the product group shows that something moved over time, even when it comes from a randomized trial.

The FTC looks for differences between the treatment and control groups. When reading clinical trial data, check whether each p-value reflects a difference between groups or a change within one group.

Mechanism and anecdote do not carry a human claim

Animal and cell studies can point to a plausible mechanism. The FTC treats them as supporting background, because results in animals or isolated cells translate poorly to benefits in people, and it expects confirmation in human RCTs.

Consumer experience sits lower still. The guidance says testimonials and consumer surveys never substantiate a health effect, because a sincere report can still come from placebo or from something unrelated to the product.

No minimum number of studies, but quality comes first

The FTC sets no required number of trials. Replication by an independent team adds weight, since a second trial lowers the odds that one result came from a bias nobody caught.

Volume does not rescue weak design. Several flawed studies do not add up to competent and reliable evidence, and the guidance says plainly that study quality outweighs study count.

The protocol details the FTC also checks

Beyond the five core principles, the guidance looks at how a study was planned and run.

• A written protocol with outcomes set in advance. The research question, the methods, and the primary and secondary outcomes are defined before the study starts, and independently validated measures carry more weight.
• Ethics review and registration. The FTC describes Institutional Review Board (IRB) review and public trial registration as generally accepted practice, and notes that registration guards against reporting only the favorable results.
• The right population. Inclusion and exclusion criteria should match the people the product is sold to.
• Every participant counted. Dropout and non-compliance are assessed, and an intent-to-treat analysis includes participants in their originally assigned groups, even if they stop treatment or don't follow the protocol. Missing data must also be appropriately addressed.

See what a full clinical study covers

A full study covers the protocol, the primary endpoint, the statistical analysis plan, IRB submission, day-to-day operations, and the final study report. Compare that scope with what your claim needs before you commit. Explore Clinical Research ▸

Where substantiation falls apart

The same five failures show up across product categories.

1. Borrowed ingredient studies

A published trial on an ingredient supports your claim only as far as it matches what you sell and what you say. The FTC asks whether the dose, the formulation, the combination of ingredients, the way the product is taken, and the study population match the advertised product.

Change any of those and a gap opens between the paper and your label. The guidance gives the example of a drink marketed as containing a "clinically tested ingredient." If the drink contains other ingredients, experts may expect testing of the finished drink.

2. Mechanism standing in for outcome

A pathway shown in cells, or an effect shown in rodents, explains why something might work. Human evidence is what shows it did, and the guidance draws that line clearly.

3. Cohorts too small to detect the effect

An underpowered study that finds nothing tells you very little either way. You have paid for an absence of information. Sample size has to be planned against the effect you expect on the endpoint you chose.

4. Endpoints chosen after the data lands

The FTC treats post hoc analysis that departs from the protocol, such as slicing the data into subgroups until one moves, as a warning sign of data mining. Such an analysis can suggest what to test next. On its own, it generally cannot substantiate a claim.

Fixing the primary endpoint and the analysis plan before enrollment is what separates a result from a coincidence.

5. Qualifiers used as a patch

Adding "may" or "helps" does not fix an unsupported claim. The FTC also cautions that words such as "preliminary" and "pilot" may sound reassuring to consumers without clearly explaining the evidence's limitations.

Calling open-label results "pilot data" does not change what the study can support. Likewise, "clinically proven" and "clinically tested" must accurately reflect the research. When the evidence falls short, work with counsel to narrow or remove the claim.

Match the study to the claim

Two questions settle the choice: how confident are you that the effect is real, and what is the evidence for?

Run a pilot study when the formulation is new, you do not know whether your endpoint can move in the window you have, or you need internal evidence before funding a larger trial. A pilot answers a go or no-go question, and the case for running a pilot first rests on that decision. A single-arm pilot tracks how participants change over time, and on its own it cannot tell you whether the product caused that change.

Go straight to a randomized, controlled study when you have a defined health-benefit claim to evaluate, enough information to plan the sample size, and a goal of substantiation. Have adequate evidence in place before using the claim in advertising.

Plan for outside readers when you are building an ingredient dossier for brands that will formulate with it. Their regulatory teams will read the methods before the results, so a randomized, placebo-controlled design with a prespecified analysis plan and public registration is the version worth commissioning.

Wait on both when your team cannot yet agree on the claim. Define the research question first, then choose the study design.

Whichever path you choose, define the primary endpoint, sample size rationale, and analysis plan before enrollment. Agree in advance on how the findings will inform your next decision.

Measure the claim from more than one angle

A claim is far harder to dismiss when several kinds of measurement point the same way.

• Wearable physiology. Consumer wearables such as Oura, WHOOP, Apple Watch, Garmin, and Fitbit record what the body did, night after night, in participants' normal routines.
• Biomarkers. Blood biomarkers and methylation-based measures such as epigenetic age can help assess biological changes relevant to the study.
• Validated instruments. Surveys such as the Pittsburgh Sleep Quality Index (PSQI) and the Patient-Reported Outcomes Measurement Information System (PROMIS) record what participants experienced.

How you use wearable data matters as much as collecting it. Reputable's research team builds endpoints from raw measures such as heart rate variability (HRV) and sleep latency, compared with each participant's own baseline, instead of from a device's proprietary readiness or recovery score.

For a sense of how those pieces fit, one of the case studies on Reputable's site, a functional coffee study, describes a 28-day randomized, double-blind, placebo-controlled trial in 70 healthy adults, with an Oura Ring tracking sleep and heart data and PROMIS surveys at baseline and day 28. You can review what a study can measure in more detail before you scope your own.

What a well-designed study still cannot do

A positive finding is not guaranteed. A well-run study can come back flat, and that result belongs in your records as much as a favorable one.

A study supports the claim it tested. A trial built around sleep quality does not substantiate a stress claim, and a result in healthy adults does not stretch to other groups.

One trial is rarely the end of the story. The FTC weighs the whole body of evidence, so a stronger set of contrary studies can still undercut a claim that your own trial supports.

Where Reputable fits

Reputable Health is a decentralized CRO (contract research organization) that designs and runs clinical studies for health and wellness brands and ingredient manufacturers. We conduct both types of studies described in this guide, so consider our perspective in that context.

• Two offerings. A pilot study and a full clinical study, and the scope follows your question.
• The pilot limit. Whether a pilot can support a claim depends on its design and findings.
• The decentralized model. Remote participation and data collection, with study design, timelines, and costs tailored to the research question.
• Ethics review. Studies go through independent Institutional Review Board review, and the review pathway follows the study design.
• Registration. Address public trial registration before enrollment, including requirements for any intended journal submission.
• Claim language for counsel. Claim language guidance shows which claims the findings can support, written for your counsel to review.
• Scoped to you. Every study is scoped to your product or ingredient, your claims, and your budget.

Common questions

Short answers to what supplement and ingredient teams usually raise once a claim is on the table.

What does "competent and reliable scientific evidence" mean?

The phrase is the FTC's substantiation standard for health claims: research designed and assessed objectively by qualified experts, using methods the field accepts as reliable. For health benefit claims, the FTC's 2022 guidance says that generally means randomized, controlled human trials.

Do I need a clinical trial to make a supplement claim?

For health-benefit claims in advertising, generally yes: the FTC expects randomized, controlled human evidence. However, existing clinical research may be sufficient if it adequately supports your specific claim. Have a qualified scientific expert and regulatory counsel assess whether that evidence applies to your formulation, dose, and intended population.

Can I use a study on the ingredient instead of my finished product?

Sometimes, if the ingredient, dose, form, delivery, and population match what you sell and what you claim. The more your product differs from the one studied, the weaker that support becomes, and a multi-ingredient product may need its own trial.

What is the difference between a structure/function claim and a disease claim?

A structure/function claim describes how a product affects the way the healthy body normally works, such as supporting sleep quality in healthy adults. A disease claim says the product diagnoses, treats, cures, mitigates, or prevents a disease, and only a drug may make one.

How long does it take to get evidence behind a claim?

Timing depends on the study design, how long outcomes need to be measured, participant recruitment, and the time needed for IRB review, analysis, and reporting. Work backward from your target launch date to allow enough time for the research and evidence review.

The bottom line

A supplement claim must be both permissible and supported by evidence. FDA primarily oversees labeling, while the FTC primarily oversees advertising. Both require claims to be truthful, adequately substantiated, and not misleading.

The evidence half is a design problem, and you can solve it before a single participant enrolls. Before your next label or listing review, have your team write the claim on one line, list every study behind it, and check each one against the FTC's design principles and the product-match questions above.