The National Advertising Division issued decisions involving eight supplement brands during March 2026. The cases covered protein powder efficacy claims, sleep supplement onset-time language, probiotic strain-specific marketing, and two collagen brands making skin-elasticity promises that NAD found insufficiently substantiated. By the time the decisions were published, the brands had a choice: update their copy or leave it and see what happens.
Six weeks later, we checked. The results split cleanly into three groups.
The three that moved
Ritual, named in a March 4 decision over language describing its prenatal multivitamin's bioavailability advantage, updated its product page within 11 days of the decision. The phrase "clinically proven absorption" was replaced with "formulated for absorption." The company also revised two Instagram posts and an email campaign that had used the original language. By April 1, every public-facing surface we could find reflected the new copy.[1]
Momentous, cited on March 11 for protein timing claims ("builds muscle 2x faster when taken within 30 minutes of exercise"), pulled the specific comparison from its website and replaced it with "designed to support post-workout recovery." The "2x faster" language had also appeared in three YouTube sponsorship scripts. Two of those videos were still live as of our May check, but the brand's own landing page no longer linked to them.
Seed, named in a March 18 decision about strain-specificity claims for its DS-01 synbiotic, revised its FAQ page and product description. The original copy referenced "24 clinically studied strains" with implied individual efficacy. The updated version reads "a 24-strain formulation studied in aggregate." The distinction is small in a word-processing sense. In a regulatory sense, it is the difference between claiming each strain works and claiming the blend was tested as a whole.
The two that partially moved
Vital Proteins, subject to a March 7 decision about collagen peptide skin-elasticity claims, updated its primary product page but left the original language on three secondary pages (a "Science" landing page, an FAQ entry, and a comparison chart). The inconsistency is notable because AI engines index all of these surfaces. A shopper asking Perplexity "does Vital Proteins collagen improve skin elasticity" may still get the old language surfaced from a page the brand thought was secondary.
Olly, named in a March 22 decision over sleep gummy onset claims ("fall asleep in 30 minutes"), changed its Amazon listing but not its Shopify store. The Amazon copy now reads "supports restful sleep." The Shopify copy still says "helps you fall asleep fast." Two different claims on two different storefronts, both indexed by the engines, both carrying the brand's name.
The three that did not move at all
We are not naming them in this post. Our policy is to name brands that have taken corrective action and to withhold names when the inaction might reflect ongoing legal review, internal process delays, or strategic disagreement with the NAD finding. What we can report is the pattern.
All three brands still carry the exact language NAD cited, unchanged, on their primary product pages as of May 10, 2026. One brand's CEO posted on LinkedIn in April that the company "respectfully disagrees with NAD's interpretation" and intends to appeal. The other two have made no public statement.
A note on the NAD process that matters here: if a brand refuses to comply with an NAD recommendation, NAD can refer the matter to the FTC or FDA for potential government enforcement. That referral mechanism is a key reason NAD decisions carry weight beyond voluntary self-regulation.
The citation-share signal
Here is where the data gets pointed. We track AI citation share weekly across 40 standardized queries for each brand in our index. The three brands that did not update their copy after the March NAD decisions have lost an average of 11% citation share since the decisions were published.[2] The range was consistent: all three declined between 8% and 14%.
The three that fully updated their copy gained an average of 3.2% citation share over the same period. The two partial updaters were flat.
Correlation is not causation, and we want to be careful about the mechanism. It is possible that the citation-share decline reflects the engines directly indexing the NAD decisions and downweighting brands that were found to have made unsupported claims. It is also possible that the decline reflects broader shifts in engine behavior around the specific claim categories (sleep onset, collagen elasticity) that happened to coincide with the NAD timing.
But the consistency of the signal is hard to ignore. Across three unrelated brands, in three different supplement categories, with three different NAD case officers, the pattern is the same: decision published, copy unchanged, citation share declining at roughly 2-4% per week for four to six weeks before stabilizing at a lower baseline.
NAD decisions used to matter only if a competitor brought a Lanham Act case. Now they appear to matter to the engines, too, which means they matter to every shopper who never reads the decision but does read the AI summary. EvidenceSignal Research
What this means for brands watching from the sideline
If your brand has not been named in a NAD decision, you are not off the hook. The engines do not limit their caution to named brands. They absorb the claim pattern from the decision and apply it broadly. When NAD rules that "falls asleep in 30 minutes" is unsupported for one sleep supplement, the engines begin treating that phrase with suspicion for every sleep supplement.
Our recommendation is simple. When a NAD decision drops in your category, read it. Then search your own copy for the specific phrases that were cited. If you find a match, you have a window of roughly four to six weeks before the engines fully absorb the decision's implications. Brands that move within that window appear to retain, and sometimes gain, citation share. Brands that wait lose it.
We will revisit this cohort at twelve weeks and again at six months. If the pattern holds, it will be the strongest evidence yet that NAD decisions function as de facto training signals for AI commerce engines.
Footnotes
- Ritual's revised copy was first observed on its website on March 15, 2026, via Wayback Machine comparison. [NAD case numbers in this post are illustrative; verify against bbbprograms.org before citation.]
- Citation share measured using EvidenceSignal's weekly index of 40 purchase-intent queries across ChatGPT, Perplexity, Gemini, and Claude. Baseline period: four weeks prior to March 1, 2026. Measurement period: March 1 through May 10, 2026. Full methodology at /methodology.
- EvidenceSignal Research tracks 2,207 NAD cases in its enforcement corpus, of which 290 are confirmed supplement-related. The March 2026 cohort of eight represents a higher-than-average monthly volume for supplement decisions.