What the labels actually say

Men's formulas versus women's formulas: the 14 comparisons #

Zero of fourteen men's or default formulas contain a probiotic strain, identified by designator, that was selected on male-specific evidence. We ran fourteen same-brand men's-versus-women's label diffs. Here is what they actually differ by — and every count below is published with the rule it applies, so a reader can check the classification pair by pair rather than take the number.

One of the fourteen does contain a male-exclusive organism, and it is the one the brand's claim points at. Garden of Life RAW Probiotics Men carries Lactobacillus fermentum and the women's SKU does not; the women's SKU carries L. reuteri and L. rhamnosus and the men's does not. Everything else matches: 22 identical kefir-culture organisms, identical Fruit & Veggie (207 mg) and Enzyme (56 mg) blends, 85 billion cultures at end of Best Use Date on both, near-identical vitamin and mineral panels (DSLD 321381 (opens in a new tab) versus DSLD 321382 (opens in a new tab), both on-market, both entered 2025-01-23). The claim and the formula are internally consistent — and still unreachable, because the label names no strain. The male-relevant L. fermentum literature belongs to a specific named strain, VRI-003 (PCC), tested in 20 elite male distance runners at 1.26×10¹⁰ CFU/day, where every primary immune endpoint was non-significant (salivary IgA/IgA1, IL-4, IL-12; IFN-γ p=0.07; illness severity p=0.06) and the only significant result was respiratory symptom days, 30 versus 72, p<0.001 (PMID 18272539 (opens in a new tab)).

In eight of the fourteen pairs the sex-specific differentiator is a botanical or a mineral. Codeage runs the same 5-strain, 50-billion chassis for both sexes and swaps one botanical — saw palmetto 80 mg for men, cranberry for women — at an identical $29.99. Physician's Choice runs the structurally identical move with selenium + hawthorn for men, D-mannose + cranberry for women. That is a cross-brand pattern, not an isolated case. In the other six the sex-specific difference is which bacteria are in the bottle — and in five of those six the brand offers no male rationale for a single one of them.

Thirteen of the fourteen women's formulas name a sex-specific anatomical target — vaginal or urinary tract — and include organisms or actives selected against it. The men's side has no anatomical target it can name and support. The one exception fails on the second half rather than the first: Ancient Nutrition's women's page does claim vaginal and urogenital support, and then attaches that claim to an undesignated Bacillus coagulans that sits in the men's bottle too, in the same blend, at the same dose.

In five of fourteen, the women's SKU is better-specified, cheaper, or both — reading "better-specified" as the panel that discloses more, never as the panel with more in it. In seven pairs the recorded evidence puts neither side ahead. In two the men's or default side leads: Hyperbiotics, where the prices are identical and the men's panel designates an L. plantarum the women's leaves at species level, and Wholesome Wellness, where the default SKU is the cheaper one — which is a 29.6% premium on the women's, not a formulation advantage for men.

And the irony that survives every check: the one organism in this category with a testosterone literature — L. reuteri — is in Garden of Life's women's products and absent from both of its men's products. That literature is five to ten mice per group.

One denominator caveat. These fourteen diffs mix men's-marketed SKUs with ungendered default SKUs from brands that sell a women's product and no men's. The men's-marketed universe is 10. Separately, 17 of the 27 products (63.0%) are ungendered SKUs relabelled "men's" by retailers and affiliate listicles — and the manufacturers say so themselves: one brand's own page reads "These mens and womens probiotics work in both the small and large intestines…", and another's section heading is "Best Probiotics for Men and Women." Four brands sell no gendered probiotic in either direction at all: Pendulum (0 of 66 products), Just Thrive (0 of 25), Transparent Labs (0 of 70), and Nature's Bounty (0 probiotics gendered across 186 products) — the last from a company that freely sex-segments its multivitamins and sells menopause products.

What the labels actually say #

We took 27 currently-sold US products apart at label level, using the printed panel, the brand's own page, and the US government's own label database. Here is what nobody else publishes.

Two denominators, stated once and never rounded away: 27 products taken apart at label level (the denominator for every product-level percentage); 294 organism-slots, being the sum of every organism named on those 27 labels (the denominator for every strain-level percentage). Both are frozen. The universe they were drawn from is a coverage figure, not a denominator, and it is counted from the collection rather than stated here.

Strain designators #

Fourteen of 27 products (51.9%) print a strain code for every organism. Five of 27 (18.5%) print codes that lead anywhere. Nine of 27 (33.3%) print none at all.

There are three distinct failure modes hiding inside "has a designator", and only the first is widely understood.

1. Species-only (9 of 27) — structurally unanswerable. A species name is not a claim.
2. The untraceable house code — Renew Life's 31 UF- codes, Seed's 24 SD-/HRVD- codes, BiOptimizers' "OM", Pendulum's WB-STR-####. All return zero PubMed hits, and all of them look exactly like an ATCC or DSM accession to a shopper who has correctly learned to "look for the strain code." Thirty of Renew Life's 31 UF- codes return zero PubMed records; the exception is GG.
3. The fabricated-looking designator. 1MD's 2021 and 2022 government label records carry, as ingredient rows: L. acidophilus L61A, B. lactis B42L, B. longum B12L, B. bifidum B66B, L. plantarum L23P, L. bulgaricus L88B, L. casei L30C, L. salivarius L10S, L. gasseri L22G, L. rhamnosus L08R. The scheme is [genus initial][two digits][species initial], constructed from the species name. All ten return 0 PubMed hits. The 2024 record carries the same ten organisms as bare species names (DSLD 246726 (opens in a new tab), 274796 (opens in a new tab), 309806 (opens in a new tab)).

43 of 294 organism-slots (14.6%) carry a strain with any human RCT, and one of the 43 rests on a resolved label typo — Transparent Labs prints LLP-115, recorded as the strain marketed elsewhere as Lp-115. Fourteen of the 27 products contribute zero. Organisms administered alone in at least one human RCT, i.e. strictly attributable: 26 of 294 (8.8%). Organisms with a human RCT for any male-specific outcome: 0 of 294 (0.0%).

Products with a published human RCT of the finished formula: 2 of 27 (7.4%). Both are named, because a count nobody can enumerate is a count nobody can check: Seed DS-01 — three sponsor-affiliated trials of the finished synbiotic, the largest an n=350 six-week efficacy trial — and Pendulum Glucose Control (NCT03893422 (opens in a new tab)), a registered multicentre trial of its own five-strain formulation. Pendulum's trial ran at three times the dose printed on its label, so on the narrower question — tested as the formula and at the dose actually sold — the count is 1 of 27, and it is Seed. 25 of 27 (92.6%) have no trial of the thing in the bottle. Every other trial registration in this cohort belongs to an ingredient rather than to the product: Legion Biome cites the B420 and xylooligosaccharide trials (NCT01978691 (opens in a new tab), NCT01944904 (opens in a new tab)), Physician's Choice cites SELECT (NCT00006392 (opens in a new tab)) for its selenium, and Just Thrive's "clinically proven" claim rests on a study whose methods name a different company's five-strain product.

The strain-designator false-positive trap is real, and it is why a bare code lookup is not enough. "Lb-87" returns a workplace weight-gain prevention programme. "Bl-05" returns B. licheniformis in a carp, duck-egg-white mayonnaise, an encapsulation study and an in-vitro growth study. "Lp-115" returns a mouse colitis model. "St-21"[tiab] AND randomized controlled trial[pt] returns 20 records, all acupuncture trials — ST-21 is the acupoint Liangmen. A bare designator query returns noise; only reading every abstract resolves it.

And the one-character trap. Hyperbiotics PRO-Men contains L. plantarum 6595. The 1996 source paper establishes the identity: "Two Lactobacillus plantarum strains of human intestinal origin, strains 299 (= DSM 6595) and 299v (= DSM 9843)…" — 299 and 299v are different strains, and 299v is the one with the consumer IBS literature (34 RCT-tagged records). The 299 trials that do exist are infection-prevention and colonisation endpoints in intubated, post-surgical and pancreatitis patients; none is an immune-health trial in healthy men, none is urological ("Lactobacillus plantarum 299" AND (prostate OR urinary OR "benign prostatic") = 0), and none is at this product's dose. The label is technically accurate and practically misleading (PMID 8779562 (opens in a new tab)).

CFU, potency and the number that isn't a number #

Only 9 of 27 products (33.3%) guarantee CFU through expiry on the label. Another 8 of 27 (29.6%) state no potency basis at all. Five state manufacture-only, two state through-expiry off-label only, one contradicts itself between label and website, and two have no CFU number anywhere.

The best potency disclosure in the frozen founding cohort of 27, and the deletion sitting next to it. Hyperbiotics PRO-Men prints both endpoints on the panel: "5 billion CFU (At the time of manufacture. Minimum of 1.5 billion CFU at expiration.)" — a 70% decline over shelf life, on that product's own label. Only one brand in 27 prints both endpoints, so no industry-wide decay rate can be inferred from it, and the direct counterexample is in the same market: Garden of Life Dr. Formulated Once Daily Men's declares 50 billion CFU at expiration with no manufacture figure at all, so its decay magnitude is unknowable, not 70%.

Why a milligram figure tells you nothing, in NIH's own words: US rules "only require manufacturers to list the total weight of the microorganisms… this cellular mass can consist of both live and dead microorganisms and, therefore, has no relationship with the number of viable microorganisms in the product."

Sixteen of the 27 products exceed 10 billion CFU, against ISAPP's own statement that "Most probiotics have been tested at levels between 1-10 billion CFU/d" and that "Simply having lots of strains is not a guarantee of a more beneficial product."

Seven documented label regressions #

Seven products removed a disclosure rather than improving it. Every one was found by comparing a current label against an older government record.

1. Nature's Bounty removed its CFU basis — archived labels said "20 billion active cultures per serving guaranteed at time of manufacture"; the current label states no basis at all.
2. Wholesome Wellness deleted its through-expiry language. The 2020 record: "Contains a minimum of 100 billion live bacteria when manufactured, and provides an effective level until at least the 'best by' date" plus "Potency guaranteed." The 2024 record: "100 billion CFU at time of manufacture." Both prior statements are gone (DSLD 240395 (opens in a new tab) → DSLD 322303 (opens in a new tab)).
3. Hyperbiotics deleted the 4-tablet dose disclosure while keeping — and promoting — the prostate claim. The 2022 record for the same UPC printed "Claim based on 4 tablets per day, providing 1000mg of Meriva Curcumin Phytosome" and a two-tier regimen. The 2025 record for the same UPC deletes that sentence, reduces suggested use to "Take 1 tablet once per day", and changes the target group from Adult Male (18-50 Years) to Adult (18-50 Years). At one tablet a day the product delivers 250 mg of Meriva — 25% of the 1,000 mg/day used in the study it evokes (DSLD 270251 (opens in a new tab), 328456 (opens in a new tab), 328457 (opens in a new tab)).
4. NOW silently swapped B. breve Bb-03 → Bb-18: same product name, same CFU, no notice. "Bb-18"[tiab] returns 0 PubMed records of any kind. "Bb-03"[tiab] returns 1, and it is a paper about UV-B tolerance in an entomopathogenic fungus (DSLD 73822 (opens in a new tab)).
5. Thorne removed per-strain amounts in 2014; the 2012 record published 2,500,000,000 / 1,250,000,000 / 1,250,000,000, and no version since does.
6. Just Thrive cut blend mass 65% (500 mg → 175 mg) while holding the stated cell count constant at 3 billion (DSLD 201373 (opens in a new tab) → current label).
7. 1MD deleted its ten fabricated-looking strain designators, leaving species-only.

The pricing finding you can check in thirty seconds #

Garden of Life's own archived RAW Probiotics Men page advertises "$47.99" and "$0.53/serving" side by side, while the government label record for the same UPC confirms 90 capsules, 30 servings, 3 capsules per serving ("Adults take 3 capsules per day, best taken as divided servings"). The advertised "per-serving" price is the per-capsule price. The real cost is $1.60/day, understated 3× (DSLD 321381 (opens in a new tab); Wayback capture 20260517144303). The control test holds: the same brand's Once Daily SKU (1 capsule = 1 serving) shows an arithmetically correct $1.23/serving.

Median verified cost across the set is $1.52 a day — about $553 a year. Both figures come off one median. Twenty-four of the 27 carry a verifiable US price, an even count, so the median is the mean of the two middle products: $1.515 a day. The daily figure is that rounded to the cent; the annual figure is $1.515 × 365 = $552.975, rounded to the dollar. Annualising the rounded $1.52 gives $554.80 instead, so the rule is printed here rather than left to the reader to guess at. And price does not buy disclosure. The cheapest verified product in the 27, at $0.38 a day, prints no CFU number at all, and the most expensive, at $3.30 a day, states no potency basis at all; the one product that prints both endpoints sits mid-table at $1.00. The product delivering 120 billion CFU per dollar-per-day has 0 of 34 designators; the product delivering 6.0 is the only one that publishes both potency endpoints. Ranking this category on CFU-per-dollar produces the exact inverse of ranking it on disclosure quality.

Seven products whose headline claim rests on a different dose than the label instructs #

Seven products whose headline claim rests on a different dose than the label instructs: the dose in the trial, the dose on the panel, and the multiple between them. Default order: as published.

Product	Dose in the trial the claim rests on	Dose the label instructs	Gap
Pendulum Glucose Control	6 capsules/day ("three capsules two times a day")	2 capsules/day	3×
Hyperbiotics PRO-Men (Meriva)	1,000 mg/day — its own deleted label sentence	250 mg/day at 1 tablet	4×
Codeage (saw palmetto)	320 mg/day — and 320 mg already failed in Cochrane and CAMUS	80 mg	4×
Legion (PreticX XOS)	1.4 g and 2.8 g in one cited trial; 2 g in the other	1,000 mg	below both
Thorne FloraMend	3 billion CFU/day, dosed twice daily in the trials	5 billion CFU once daily	schedule differs
Transparent Labs (Bl-04)	20 billion	10 billion	2×
Seed (SD-LPLDL-UK)	4×10⁹ CFU/day	the entire Cardiovascular blend is 4.21 billion AFU shared between two organisms	below

Replicating the Pendulum trial regimen costs $297–417 per month — $9.90 to $13.90 a day.

Copy that does not match the panel #

Renew Life's product page credits twelve branded strain designators with benefits, and only one of them is in the bottle. The live benefit table names GG · Bi-26 · HN019 · La-14 · Bi-07 · BI-04 · NCFM® · Lpc-37® · LBr-35™ · Lc-11® · Li-23™ · Lp-115®. Only L. rhamnosus GG appears on the Supplement Facts panel. Eleven of twelve are absent. Narrower and stronger: of the eight designators credited with a health benefit, seven are not in the product. The same page states both "12 probiotic strains" and "90 Billion CFU from 31 probiotic strains", and the panel has 31. The mechanism is visible in the near-miss numeral collisions — UF-35/LBr-35, UF-23/Li-23, UF-115/Lp-115. The most likely explanation is un-updated page copy after a reformulation, and we are not alleging intent.

Two products publish no Supplement Facts panel in any form. Onnit Total GUT HEALTH: a token census on both live product pages returns Supplement Facts 0, CFU 0, Servings Per 0, Bacillus 0, third-party 0, NSF 0 — and the 60-count product is a 15-day supply at four capsules per serving. Seed DS-01 publishes four blend totals with no "Supplement Facts" heading, no serving size, no servings per container, no %DV column and no Other Ingredients list.

Other things found on labels or brand pages, verbatim. A brand's own panel prints "Bifidobacterium brevis" (an invalid name) and counts Saccharomyces boulardii — a yeast — inside a stated total of "13 Strains of Probiotic Bacteria", while the bottle says 1 capsule/day and the brand's own page says "Take 2 capsules daily." Another prints "Organic Apple Fiber (Mallus pumila) Pomace" — the genus is Malus. One label and its own website contradict each other on both potency basis and storage: the label says "60 billion CFU at time of manufacture… Best stored refrigerated", the site says "no refrigeration required… guaranteed through the expiration date, not the manufacturing date." And a patent number printed on an FDA-filed label resolves to an accelerometer: BiOptimizers' 2024 record prints "patent # US 5,895,858 A", which is "Integrated accelerometer test system", assignee Analog Devices Inc — 278 occurrences of "accelerometer", zero of "plantarum." The real strain patent, US 5,895,758, is Expired – Lifetime with anticipated expiry 2017-06-10, while the page markets a "New patented strain" whose sole active has no human evidence of any kind (P3-OM → 0 PubMed records; "plantarum OM" on ClinicalTrials.gov → 0 studies).

What "reviews" and "guarantees" are worth here #

One brand's efficacy percentages come from compensated customer surveys, and the brand says so in its own footnotes: "†Based on a 2021 survey of 312 Complete Probiotics Platinum® repeat customers. Customers were compensated for their survey participation" — supporting the claims "Month 3: 73% of users feel less gas", "4 out of 5 users notice they are more regular" and "82% of users report significantly less bloating." The FTC's position on that category of evidence is one sentence: "Consumer endorsements themselves are not competent and reliable scientific evidence" (16 CFR §255.2(a)).

One brand's 365-day guarantee voids on the second bottle, verbatim: "…just send back your remaining unopened bottles… If you open a second bottle, you no longer qualify for any guarantee." For a 30-count daily product, that is a 30-to-60-day guarantee.

One six-review, 5.0-star on-site rating includes unmoderated spam. The brand's own schema.org block reads "aggregateRating": {"ratingValue": 5.0, "reviewCount": 6}, and the most recent entry is a 5-star review from a "Verified Reviewer" whose body advertises a herbal herpes cure with a gmail address — counted in the average.

And Amazon's displayed review counts are inflated by variation-merging, which is mechanically demonstrable. A men's probiotic listing embeds reviews tagged data-asin="B0FH61LCW8" while the page's own twister state reads {"asinsInCollapsedView":["B0FH61LCW8","B0D41MW6T2"]} — the first being the 180-count variant of the same product. That is a verifiable inflation vector, distinct from fake reviews.

The label-accuracy literature, both halves #

Label accuracy is bimodal by market, and the two halves must never be averaged. In regulated Western markets: 17 of 52 US supplements (33%) were already below their label CFU claim before their expiration dates, with only 30/52 (58%) correctly labelled by classification (note: DuPont Nutrition & Health authorship); only 1 of 16 products perfectly matched its bifidobacterial label claims in all samples; 15 of 182 (8%) non-compliant across the US and Canada; only two of ten Bacillus formulations matched their labels qualitatively and quantitatively, with B. cereus among the contaminants. In weakly regulated markets the picture is far worse — "Almost none of the probiotic products tested met the label claim" (Pakistan); 29 of 45 (64.4%) incorrectly labelled (Thailand).

The honest counterweight comes from FDA's own laboratory: "Overall, the products we tested adhered to the ingredient claims and lot-to-lot differences were minimal", with a few species-naming discrepancies, one apparent Enterococcus contaminant, and "Of all the supplements examined, only one was found to be inaccurate in viability" (PMID 27303722 (opens in a new tab), FDA CFSAN, Laurel MD). The literature is not unanimously damning.

Misidentification is industry-wide and decades old: 213 cultures from 26 manufacturers and four collections produced 46 misidentifications, including 34 commercial probiotic cultures deposited by 10 different companies — "more than 28% of the commercial cultures intended for human and/or animal probiotic use were misidentified at the genus or species level" (PMID 16919915 (opens in a new tab)). And even a competent lab often cannot resolve a strain: in one study targeted PCR confirmed every species, but only 40 of 100 strains could be identified to strain level, because strain-specific methods do not exist for the other 60 (PMID 37138639 (opens in a new tab)).

Three caveats ship with any label-accuracy number. DNA is not a viable organism. The answer depends on which 16S region you amplify. And plate counts are noisy and undercount, with ±0.5 log₁₀ as a practical acceptance window. The best illustration that statistical significance is not clinical meaning: 12 UK products all deviated significantly from label claims (p<0.05), yet 11 of 12 met the ±0.5 log₁₀ benchmark — while one human product contained no detectable viable bacteria (PMID 40871437 (opens in a new tab)).

Storage matters, with measured magnitudes. After 90 days: 4.2±0.1 to 5.3±0.2 log CFU/g refrigerated versus 2.6±0.3 to 4.9±0.2 log CFU/g at room temperature. In three commercial probiotic yoghurts, "only 53.8% of the CFU remained after 6 h at RT; after 24 h, only about one fourth of the CFU were found."

Machine-readability, and what it costs. Fifteen of 27 products (55.6%) publish no machine-readable Supplement Facts panel. A screen-reader user cannot access the ingredient list for those 15. Price-comparison tools, allergen checkers, drug-interaction checkers and AI shopping assistants cannot parse them. Where a formula changes there is no public record of what it used to be, except where the government database happens to hold a superseded copy — which is exactly how the seven label regressions above were found. And three brands hard-block automated retrieval of their own product pages, so a consumer or a tool cannot verify the label against the manufacturer's own site at all.

Certification: what the issuers' own registers actually say #

We queried every certifier's own public database rather than trusting a badge. Only 2 of 27 products (7.4%) hold a certification we could confirm at the issuer. Those two are Legion (Labdoor, score 100.0, lot 0385A6, tested 14 July 2026) and Sports Research (Non-GMO Project, brand id 9714, product 2-25768-3023, status Verified, confirmed via the Project's own open API).

USP Verified. The entire USP Verified probiotic register contains exactly five products, two of them explicitly women's and none men's: Culturelle Digestive Daily · trunature Women's Daily · trunature Advanced Digestive · Member's Mark 10 Strain · Member's Mark Women's Daily. Against a programme stating that "More than 160 different dietary supplement formulas have received the USP Verified Mark", that is roughly 3%. Zero men's-marketed probiotics hold it. And USP says in writing what the mark is not: "USP's Verification Program does not comprehensively address the issue of efficacy of a dietary supplement." It verifies identity, quantity and contaminant limits. Note also that company participation is not product verification — USP's participants page lists brands whose products do not appear on the verified probiotic list. One brand holds the mark and does not display it: Culturelle's page shows four other badges and no USP mark, while the product sits on USP's own register.

NSF. Zero of the 27 hold NSF Certified for Sport, a programme that tests for 290 banned substances (per its get-certified page; the homepage says "280+" — cite 290 and note the discrepancy) with lot-level testing. But the correct men's finding is not zero: at least two men's-marketed probiotic-containing products hold a real, register-checkable NSF/ANSI 173 listing — Garden of Life Dr. Formulated Probiotics Mens Daily Care (facility Bohemia NY; a different SKU from Once Daily Men's, which holds nothing) and SmartyPants Men's Organic Multi & Probiotics (facility Acton Vale, Québec). The NSF/ANSI 173 register returns, verbatim, "Number of matching Manufacturers is 39 — Number of matching Products is 160" for TradeName=probiotic — roughly ten times the Certified for Sport probiotic count and thirty times the USP count. Designs for Health "Probiotic 50" is the only entry in that register whose category tag genuinely reads Men/Probiotic/Women.

ConsumerLab. Its CL-Certified probiotic list contains exactly nine products: none men's, two pet, one women's. ConsumerLab publishes the strongest public pass criteria in the category — a product must "Contain a minimum of 100% of its labeled amount of microorganisms as viable anaerobic microorganisms (based on CFUs found, even if product lists amount in AFUs… or lists amount only as of time of manufacture)" and yield at least 1 billion anaerobic microorganisms per maximum recommended daily serving. Two structural caveats: the paid arm is survivorship-biased in its own words ("There is a testing fee paid by the manufacturer and results are proprietary… However, if a product 'Passes,' it may appear in CL's website listing"), and ConsumerLab does not screen most probiotics for heavy metals — its ICP-MS trigger requires whole herbs or more than 250 mg of minerals per daily dose, and a plain probiotic capsule is neither.

Labdoor. Labdoor's entire probiotics category contains exactly one product. labdoor.com/rankings/probiotics reads "TOP 1 PROBIOTICS — See our 1 product reports." Any "#1 ranked by Labdoor" framing in this category is #1 of 1. Labdoor's own conflict is on its own site: "When you shop at Labdoor, we keep about 10% of each sale", alongside an enterprise column offering "Custom Testing", "Certifications" and "Manufacturer Partners." And of Labdoor's five score components, four come from lab assay and one — Projected Efficacy — is modelled, not measured.

Clean Label Project. The Purity Award is a percentile, not a standard, and there is no probiotic category study. CLP Certified means absolute standards derived from Prop 65, EU and US federal/state rules. The Purity Award adds "recognizes only those products that perform among the top one-third of all products tested within their category", granted only in categories CLP has comprehensively studied — and CLP's published category studies are Protein Powder, Snack & Nutrition Bars, Coffee, Pediatric OTC, Infant Formula and Dog Food. No probiotic study exists.

Informed Sport versus Informed Choice are not equivalent, and LGC says so. Informed Sport (est. 2008) is "the only global third-party testing and certification programme that tests every single batch of a product prior to it being released to market." Informed Choice (est. 2007) tests "At least twelve lots/batches of the product… at random every year" via retail monitoring. Zero of the 27 hold either. And the widely repeated "one in ten supplements is contaminated" line is certifier marketing: Informed Sport states it with no underlying study cited on the page.

One finding about verification itself. Certifier registers are frequently unreachable to automated access — nsf.org main site 403, B Lab directory 403, Non-GMO Project legacy paths 404, vegan.org 403. That is a finding about the verification infrastructure, not about the brands. Where a register could not be queried, we say the claim is brand-asserted and unverified rather than calling it false.

How to check any label in five minutes #

1. Find a strain code, not a species name. You are looking for something like DSM 9843, ATCC 53103, CNCM I-745, MTCC 5260. A species name — "Lactobacillus acidophilus" — is not a claim. Three different L. acidophilus strains produced three different results in the same indication.
2. Search the exact code in PubMed, then read the abstracts. Do not trust the hit count. "St-21" returns 20 randomised trials, all about acupuncture. House codes that look like accessions (UF-, SD-, WB-STR-, OM) return nothing at all.
3. Check the potency basis, not the CFU number. "At time of manufacture" tells you what was in it at the factory. "Through expiry" or "at expiration" tells you what you are buying. Only 9 of 27 products state the second on the label. ISAPP's own criterion: "Responsible probiotic manufacturers will list the genus, species and strain of the product, as well as the potency of the product to expect through the end of its shelf life."
4. Check the dose against the trial, and the schedule too. A strain at a fifth of the studied dose is not that strain's evidence. Neither is a once-daily label backing a twice-daily trial. And a proprietary blend total tells you nothing about any single organism in it — 20 of 25 multi-organism products in our set (80.0%) conceal the per-organism split.
5. Do the per-day arithmetic yourself. Capsules per serving × servings per day. A "$0.53 per serving" claim on a three-capsule-a-day product is a $1.60-a-day product.

A sixth rule, for the men's aisle specifically: ignore the sex on the bottle. In 13 of 14 same-brand comparisons nothing on the men's or default side is a bacterium the brand offers a male rationale for — the differentiator is a botanical, a mineral, an organism swap with no rationale attached to it, or nothing at all. In the 14th, the male-exclusive organism is named only to species, so no trial can be matched to it.