Claims & Labelling
Product classification follows the whole product: the formula, the intended use, the label copy, the imagery and the advertising. A compliant formula described in therapeutic language becomes a different regulatory category, with a different approval pathway and a different set of obligations.
That reclassification happens in marketing, not in the lab, and it is usually avoidable without changing the product at all.
Common failure modes
Relief, recovery, treatment, anti-inflammatory. These communicate therapeutic intended use regardless of what is in the formula or at what level.
A footnote does not cure a misleading headline. If the main message communicates a therapeutic effect, small print below it does not reverse that.
A pharmacokinetic result from a fasting capsule study does not automatically support a claim on a finished food product. The ingredient data and the product data are two different things, and only one of them is about what you are selling.
A grade or content claim printed before finished-product testing confirms it is a compliance exposure even when the number turns out to be right.
The mechanism
A compliant formula described in therapeutic language becomes a different regulatory category, with a different approval pathway and a different set of obligations. The product does not have to change for that to happen, and it does not have to change to prevent it either.
Most of this exposure is removable at the copy stage, at a fraction of the cost of removing it after artwork is signed off.
The method
Every term you intend to use is assessed individually and marked with the reasoning behind it. Where a term fails, we supply a compliant alternative that keeps the consumer appeal.
Allowed
Allowed with qualification
Conditional
Do not use
For each claim you want to keep, we identify what evidence supports it and where the evidence stops. Claims that need finished-product work to be defensible are flagged before they reach artwork.
Ingredient declarations, mandatory statements, format requirements and the interaction between copy, imagery and prominence. Artwork is reviewed as a whole, because context changes the meaning of otherwise acceptable words.
Where a professional recommendation is used, we set out what the endorsement file has to contain and how the qualifier must be worded.
Coverage
We assess claims against the framework that will actually apply to the pack in each market you intend to sell in.
Health Canada NNHPD
FDA and DSHEA
European Union
TGA Australia
Cosmetic frameworks
BIS where relevant
SUBSTANTIATION
A claim holds up when the evidence behind it matches the product being sold: the same ingredient form, a comparable dose, a relevant population, a meaningful endpoint. Most substantiation files fail on the gap between the study and the finished formula rather than on the absence of studies. We build the file from the claim backwards, so the gap is visible before the claim is printed.
Claim-to-evidence matrix for every claim in use.
Literature review focused on the finished formula rather than the ingredient in isolation.
Assessment of dose, form and population match between study and product.
Identification of claims needing finished-product evidence.
Study planning support where existing evidence stops short.
DIGITAL AND AI-GENERATED CONTENT
Product claims now live in listings, social posts, influencer scripts, testimonials, email sequences and increasingly in copy produced by AI tools. Regulators assess the overall impression a consumer receives across all of it, not only the words on the pack. AI-generated copy is a particular risk because it writes confidently in exactly the therapeutic language supplements cannot use.
Review of websites, marketplace listings and landing pages.
Influencer briefs and scripts reviewed before content is produced.
Testimonial and review content assessed for implied claims.
An approved claims library that marketing teams and AI tools can work from.
Periodic review of live content against the approved library.
A claim matrix is fast, scoped and cheaper than a reprint. The best time to run it is while the copy is still a document, before it becomes a print file.