Claims Substantiation: The Evidence Hierarchy and How to Build a File That Holds Up

The standard: a reasonable basis, before you claim it

FTC advertising law comes down to two rules: ads must be truthful and not misleading, and advertisers must have adequate substantiation for every objective claim before they run it — not sourced afterward if someone asks. For health-related products, that substantiation has to take the form of "competent and reliable scientific evidence," which FTC defines as tests, analyses, research, or studies that have been conducted and evaluated in an objective manner by qualified experts, using procedures generally accepted in the field to yield accurate results (FTC Health Products Compliance Guidance).

There's no fixed checklist for how much evidence is "enough." The FTC weighs it on a sliding scale based on the type of product, how specific and quantified the claim is, the cost of developing better evidence, the consequences of the claim being false, and what experts in the relevant field would consider reasonable. A vague, low-stakes claim needs less support than a quantified efficacy claim ("reduces joint pain by 40% in 8 weeks") or a claim consumers can't verify for themselves. The tighter and bolder the claim, the higher the bar.

The evidence hierarchy

Not all evidence is equal, and FTC guidance is explicit about where each type ranks. Building a file without understanding this hierarchy is the single most common way brands end up with a claim they can't actually defend.

The practical takeaway for a claims file: if your strongest evidence for a specific efficacy claim is an in vitro study, a rodent model, or a stack of five-star reviews, you don't have substantiation for that claim yet — you have a hypothesis and a marketing risk.

What makes a study actually count

Having an RCT isn't automatically enough — the FTC also scrutinizes the internal validity of the study itself. When you're logging a study into a substantiation file, these are the specific factors reviewers (and competitors challenging your claims through NAD) will check:

• Control group. A treatment group alone proves nothing — improvement could be placebo effect, natural progression, or the practice effect. Studies need a placebo or sham-treatment control to isolate the product's actual effect.
• Blinding. Both subjects and researchers should be blinded to group assignment. Unblinded studies are far more vulnerable to conscious or unconscious bias in how results are measured and reported.
• Randomization. Subjects need to be randomly assigned (or carefully matched) so age, diet, health status, and other variables are balanced between groups. Imbalanced groups can produce results that have nothing to do with the product.
• Statistical significance and clinical meaning. A result has to be statistically significant to rule out chance — but a statistically significant result that's too small to matter to a consumer isn't a basis for a marketing claim. Watch for studies that measured multiple outcomes and only reported the ones that came out favorable; that's a red flag for "p-hacking."
• Sample size and duration. The study needs enough subjects and enough follow-up time to actually detect the claimed effect — and any safety signal — reliably.
• Dose and formulation match. This is where brands most often self-sabotage: if the study used a different dose, delivery form, or combination of ingredients than what's actually in your product, it doesn't transfer. A study on 500 mg of an extract doesn't substantiate a claim about your 150 mg blend.
• Population match. A trial in postmenopausal women with diagnosed joint conditions doesn't automatically substantiate a general wellness claim aimed at all adults. The tested population has to reasonably reflect who you're marketing to.
• Protocol registration and peer review. Pre-registered trial protocols, IRB oversight, and rigorous peer review all add credibility. A study that's merely "published" isn't automatically reliable — publication-fee journals with minimal review exist, and FTC explicitly notes it will demand underlying data and raw protocols for research that hasn't gone through genuine peer review (FTC).
• Intent-to-treat analysis. Results should include everyone originally assigned to a group — including dropouts and non-compliant subjects — not just the subset who finished and responded well.

Totality of the evidence — no cherry-picking

FTC guidance is direct on this point: studies can't be considered in isolation, and advertisers can't rely on the study that supports the claim while ignoring other well-conducted research that doesn't. If you have two small positive studies and one larger, better-controlled study that found no effect, the larger negative study controls — continuing to tout the earlier favorable results would itself be deceptive.

The FTC's own published example illustrates the standard clearly: a company relying on a small favorable study while a later, larger trial (including a post hoc subgroup analysis) found no significant effect is not substantiated — and a narrowly qualified claim that only cites the favorable post hoc results is still deceptive. On the other hand, a claim backed by a well-designed 12-week RCT, corroborated by multiple independent European trials using a similar dose and formulation, and confirmed by independent expert review of the full body of evidence, is the kind of file that holds up (FTC Health Products Compliance Guidance).

Practically, this means your substantiation file needs a section for contrary or inconsistent evidence, not just a folder of studies that agree with you. If contradictory research exists and you don't have a sound explanation for the inconsistency (different dose, different population, weaker methodology), the claim needs to be narrowed, qualified, or dropped.

This isn't theoretical: recent enforcement

Substantiation challenges aren't rare, and they don't only come from regulators — competitors routinely file challenges through BBB National Programs' National Advertising Division (NAD), a self-regulatory body whose decisions can be referred to FTC and FDA when a company doesn't cooperate. Two 2025 cases show both ends of the outcome spectrum:

The lesson from both cases is the same: a substantiation file isn't paperwork you assemble defensively after a challenge arrives. Companies that can pull a claim-by-claim evidence log quickly and substantively respond fare very differently than companies that can't.

Building an audit-ready substantiation file

An audit-ready file should let you answer, for any single claim on any label or ad, "what's the evidence and where is it," in minutes — not by searching through old email threads with a supplier. At minimum, each claim entry should capture:

• The exact claim text, and every channel/SKU it appears on (label, website, Amazon listing, TikTok Shop content, influencer scripts)
• Whether it's a structure/function, general well-being, nutrient deficiency, or other claim type, and its FDA notification status if applicable
• The specific ingredient, dose, and formulation the claim is about — matched against what's actually in the finished product
• Every study relied on, with citation, study design (RCT, epidemiological, animal, in vitro), sample size, duration, population, dose tested, and outcome
• An evidence-hierarchy grade for each study and an overall grade for the claim (e.g., "strong — 2 independent RCTs, matched dose and population" vs. "weak — single small unpublished pilot")
• Contrary or inconsistent evidence you're aware of, and your documented rationale for why the claim still holds despite it
• Date substantiated, date last reviewed, and who signed off internally

Keep this file current and centralized rather than scattered across supplier tech packs and old marketing decks — when a challenge or notification review arrives, the ability to produce it fast is itself part of demonstrating good faith.

When to re-open a claim you already substantiated

Substantiation isn't a one-time event — the underlying science and your own product can both move. Re-open and re-grade a claim whenever:

• The formulation, dose, ingredient source, or delivery form changes, even slightly
• New research is published on the ingredient — favorable or unfavorable — since the totality of evidence, not just your original studies, determines what's defensible
• Marketing copy is revised and the claim's wording has drifted from what the original evidence actually supports (a common and dangerous pattern: "supports healthy blood pressure" quietly becoming "lowers blood pressure" over successive rounds of ad copy)
• A competitor or NAD challenge specifically calls out the claim, even if it targets a different brand making a similar claim about the same ingredient
• It's been longer than 12 months since the last review — a periodic cadence catches drift before it becomes a pattern

Common mistakes that sink a file under review

• Treating an FDA structure/function claim notification as if it were FDA approval of the underlying science — notification is a labeling formality, not agency review
• Relying on a study of a different dose, ingredient form, or combination product than what's actually sold
• Citing only favorable studies while ignoring larger or better-controlled research that found no effect
• Using in vitro, animal, or anecdotal evidence as if it were sufficient on its own for a human health claim
• Letting marketing copy outpace the evidence over successive revisions without re-checking the file
• Having no single, centralized record of which studies support which specific claims across every sales channel

This guide is educational and reflects FTC and BBB National Programs (NAD) published guidance as of July 2026. It is not legal or regulatory advice, and does not replace review by qualified regulatory counsel for your specific products and claims.

• FTC — Health Products Compliance Guidance
• FTC — Dietary Supplements: An Advertising Guide for Industry
• BBB National Programs — National Advertising Division Finds Certain Reus Cata-Kor NAD+ Claims Supported; Recommends Others Be Modified or Discontinued
• GlobeNewswire / BBB National Programs — National Advertising Division Refers Virgin Scent to Regulatory Authorities