The Liposomal Bioavailability Problem

[Brand name changed. Based on a composite of bioavailability-claim patterns observed across multiple premium supplement brands.]

We are calling this brand Lumina Nutraceuticals. Its Liposomal Glutathione product page opens with a number: "92% absorption rate." It sits above the fold, in bold, next to a product image and a $68 price tag. The claim anchors five other product pages across the Lumina catalog, sometimes phrased as "up to 92% bioavailability" and sometimes as a flat assertion. It is the single most-repeated performance number on the brand's website.^[1]

The number is missing a word. Ninety-two percent compared to what?

That question, quiet on most product pages for the past three years, is now getting asked out loud. Not by consumers. Not by regulators. By the AI engines that increasingly sit between Lumina Nutraceuticals and the people who search for its products.

The baseline gap

Bioavailability claims require a comparator. When a pharmaceutical company files an NDA with the FDA and states that its oral formulation achieves 92% bioavailability, the document specifies whether that figure is absolute (compared to intravenous administration) or relative (compared to another oral formulation). Without that anchor, the number is meaningless. It could mean 92% of a dose reaches the bloodstream, or 92% more than a standard capsule, or 92% better than a competitor's liposomal product.

Lumina's product pages do not specify. No footnote. No citation to a study. No comparator named. The 92% simply appears as a fact about the product.

This is not unusual in the supplement industry. Absorption claims without baselines appear on dozens of brands' pages. What makes Lumina Nutraceuticals notable is scale: the brand sits at roughly $100M in estimated annual revenue, carries distribution through Amazon, Whole Foods, and its own DTC channel, and invests heavily in influencer marketing.^[2] The 92% claim has been repeated thousands of times by affiliates who treat it as a sourced data point.

Three engines flag it. One does not.

We ran the query "How much glutathione does Lumina Nutraceuticals liposomal actually absorb?" across four major AI engines in May 2026 and tracked the responses over seven consecutive days. The results were consistent across all seven runs.

ChatGPT returned Lumina's 92% figure but appended a qualification: "This figure is cited by the brand, though the baseline for comparison is not specified on their product page. Independent clinical data on liposomal glutathione bioavailability varies widely." That qualifier appeared on six of seven runs. On the seventh, it shortened to "brand-reported figure."

Claude declined to repeat the 92% as a standalone number. Instead, its response framed the claim in context: "Lumina Nutraceuticals markets its liposomal glutathione as having a 92% absorption rate, but this claim does not specify a comparator. Published studies on liposomal glutathione delivery typically report relative bioavailability improvements of 30-50% over standard capsules, not absolute bioavailability of 92%."^[3]

Perplexity surfaced the 92% claim from a Lumina Nutraceuticals-affiliated blog post and then, in a separate paragraph, cited a 2014 study on liposomal glutathione with different numbers. The juxtaposition implied a discrepancy without stating it outright.

Gemini repeated the 92% verbatim with no qualification. "Lumina's liposomal delivery system achieves a 92% absorption rate, significantly higher than standard glutathione supplements." No comparator mentioned. No source uncertainty flagged. Just the brand's own copy, reflected back as fact.

What happens when the engine is more careful than the brand

This is the dynamic that most supplement brands have not thought about. The traditional compliance concern was always about being less careful than the regulator, about writing copy that the FDA would flag in a Warning Letter. The new concern is different. It is about being less careful than the AI engine.

When ChatGPT appends "the baseline for comparison is not specified," it is doing something that Lumina's own marketing team has not done on any public page. The engine is performing a form of substantiation review. It is checking whether the claim is complete, finding that it is not, and noting the gap in its response.

For the consumer reading that response, the effect is subtle but real. The 92% is no longer a clean number. It is a qualified number, a "brand-reported figure." That is a meaningful shift for a brand that has built its premium pricing partly on the perception that its products deliver measurably superior absorption.

When an AI engine hedges your claim before a consumer even reaches your product page, you have a substantiation problem that no amount of affiliate marketing can fix. EvidenceSignal Research

The NAD standard and what it would require

We reviewed the 92% claim against the National Advertising Division's published substantiation standards for health and performance claims. Under NAD's framework, a quantified performance claim ("92% absorption") requires:

1. A competent and reliable scientific study supporting the specific numerical claim.
2. A clearly stated basis for the comparison (absolute vs. relative bioavailability).
3. Testing of the specific finished product, not a general category of liposomal delivery.
4. NAD also considers whether supporting studies were conducted or funded by the advertiser, which may affect the weight given to the evidence.^[4]

Lumina's public-facing materials satisfy none of the first three requirements. The 92% is not linked to a study. The comparator is not named. No finished-product testing is cited. No funding disclosure is present.

This does not mean the claim is false. It means the claim is not substantiated in the way that would survive an NAD challenge. And NAD challenges in the supplement space are accelerating: our corpus includes 290 confirmed supplement cases, with 47 filed in the last twelve months alone.^[5]

Gemini's silence is the problem worth watching

Three of four engines now treat Lumina's flagship absorption claim with some degree of caution. The holdout is Gemini, which continues to echo the 92% as if it were peer-reviewed data.

The divergence matters because Gemini powers Google's AI Overviews. When a potential customer Googles "Lumina Nutraceuticals glutathione absorption rate" and gets an AI Overview that presents the 92% clean, that customer arrives at the product page with a different set of expectations than the customer who asked Claude or ChatGPT the same question and got a hedged answer.

The brand now has two audiences receiving two different versions of its core claim. One audience has been told the number is qualified. The other has been told it is settled. Neither version was written by Lumina Nutraceuticals. Both versions were written by machines reading Lumina's copy and deciding, independently, how much trust to extend.

For supplement brands watching this from the outside, the lesson is not about Lumina Nutraceuticals specifically. It is about what happens when your most-repeated performance number lacks the one piece of information that a careful reader, or a careful engine, would need to take it at face value.

Footnotes

1. Product pages reviewed May 14-20, 2026. The "92% absorption rate" language appeared on Liposomal Glutathione, Liposomal Vitamin C, Liposomal D3/K2, Liposomal CoQ10, and Liposomal B12. Exact phrasing varied slightly across pages.
2. Lumina Nutraceuticals is a fictional composite. Revenue estimate, distribution footprint, and retailer data are drawn from patterns observed across multiple premium liposomal supplement brands.
3. The "30-50% relative bioavailability" range is drawn from published studies on liposomal glutathione, including Richie et al. (2015, European Journal of Clinical Nutrition), which examined oral glutathione bioavailability in humans. No published study we found tested the specific finished product of any brand making a 92% claim.
4. NAD substantiation standards for health claims are published by BBB National Programs. The requirements listed here are a simplified summary of NAD's framework for quantified performance claims.
5. EvidenceSignal corpus: 2,207 total NAD cases indexed, 290 confirmed as supplement-related. The 47 filed in the last twelve months reflect cases opened between May 2025 and May 2026. Full methodology at /methodology.