One product, one panel, one record.
Align Probiotic 70 Billion CFU Capsules — label audit
Six organisms, six designators — and four of them return zero PubMed records at their exact code, while the fifth's 322 records collapse to two human RCTs measuring tight-junction proteins rather than any outcome this product claims.
This page is an audit of a label. It records what this product’s panel discloses, what of that could be checked against a primary record, and what could not. The score below ranks this product on that disclosure and on nothing else.
The record this page rests on
- NIH label record
- No NIH Dietary Supplement Label Database record is on file for this product in this teardown.
- Label read on
- The most recent dated artefact behind this page. Labels change; this is how old our reading of it is.
- Cohort
- Addendum — outside the founding denominators Torn down 2026-07-31, after the founding cohort of 27 was frozen. It sits outside every D27 denominator the site publishes.
- How deep this went
- Full forensic teardown Panel transcribed, every strain designator queried, every certification chased to its issuer’s own register, and the literature behind each marketing claim opened and read.
- What did not survive verification
- Part of this page was refuted or could not be reproduced Two things in this record are not quotable as label facts. First, the ~1.4% share attributed to Bifidobacterium longum 35624 — the strain every research claim on the page is about — is a CROSS-LABEL INFERENCE, not a number printed on this panel. It is derived by reading the 10 mg of 35624 here against the brand's own current 24/7 Digestive Support label, which declares the same strain at the same 10 mg as 1 x 10⁹ CFU at manufacture. This panel breaks out no per-strain CFU, which is the finding itself, and it is possible Align loads a more concentrated 35624 preparation here. Publish the arithmetic and the uncertainty together or not at all. Second, no DSLD record exists for this SKU — confirmed with five query spellings — so the entire panel is single-sourced on the brand's own label image with no independent federal transcription behind it.
The Label Verifiability Score for this product
65.1 out of 100
Rank 66 of the 154 products in this audit the rubric can score.
The score measures one thing: how much of what you are buying this company tells you, and how much of that survives being checked against a source that is not the company. It does not measure whether the product works — this site holds no efficacy data of its own, and 0 of 294 organism slots in the founding cohort carry a strain with a human randomised trial for any male-specific outcome, so nobody in this category has evidence that would support a ranking by effect.
Computed over 86% of the rubric by weight — the share of the five questions below this record could answer. The remaining question is unanswered, and unanswered is not nought: it leaves the divisor rather than dragging the score down. That share travels with the number and is never dropped — this score over 86% of the weight and the same score over all of it are different claims.
| Component | Weight | What was counted, and where | Fraction | Weight × fraction |
|---|---|---|---|---|
| Strain identity printed on the panel | 0.24 | 6 of 6 organism slots carry a printed strain designator. | 6 / 6 = 1.0000 | 0.2400 |
| Per-organism amount printed | 0.20 | 1 of 6 organism slots have an amount of its own; the rest are inside a blend total. | 1 / 6 = 0.1667 | 0.033333 |
| What the CFU count is measured against | 0.22 | Potency basis "both-endpoints": CFU figure printed yes, single basis stated yes, basis reaches expiry yes — 3 of 3. | 3 / 3 = 1.0000 | 0.2200 |
| Printed designators that lead to a human trial | 0.20 | 2 of 6 checked organism slots resolve to at least one human randomised trial under the designator exactly as printed. This counts whether the identity leads to a trial, never what the trial found. | 2 / 6 = 0.3333 | 0.066667 |
| Certification confirmed at the issuer’s own register | 0.14 | No certifier register has been queried for this product. Not checked is not the same as holds nothing. | Not answered | — excluded from the divisor, not counted as nought |
0.5600 — the last column added up — divided by 0.86, the weight this record could answer, is 0.6512. Multiplied by 100 and rounded to one decimal, that is the 65.1 above.
See this product’s row in the full ranking, alongside the other 153 scored products, or read what each of the five components asks and how to check it .
What this label discloses
Everything in this table is what the product itself states — on the printed panel, on the brand’s own page, or in the label record the manufacturer filed. Nothing in it is checked yet; checking is the next section.
| Disclosure | As recorded |
|---|---|
| CFU as printed | 70 billion (70 x 10⁹ CFU) when manufactured; 15 billion (15 x 10⁹ CFU) until the "best by" date |
| What that count is measured against | Both endpoints printed |
| Strain designators | All organisms coded |
| Organism slots transcribed here | 6 — the full panel |
| Marketed as | Unisex |
"† Contains 70 billion (70 x 10⁹ CFU) probiotic cells when manufactured, and provides 15 billion (15 x 10⁹ CFU) probiotic cells until the 'best by' date."
The panel’s own words on potency, verbatim.
The panel prints both a manufacture figure and an expiration figure, so the decline over shelf life is readable off the label itself.
Every organism on the panel carries a printed strain designator. Whether those designators lead anywhere is a separate question, answered below.
| Organism as printed | Designator as printed | Quantity as printed | What the designator resolves to | The query we ran |
|---|---|---|---|---|
| Bifidobacterium longum subsp. longum | 35624™ | 10 mg | Human RCT | ("Bifidobacterium 35624"[tiab] OR "Bifidobacterium longum 35624"[tiab] OR "Bifidobacterium infantis 35624"[tiab] OR "UCC35624"[tiab] OR "UCC 35624"[tiab] OR "NCIMB 41003"[tiab]) AND randomized controlled trial[pt] AND humans[mh] → 8 |
| Lactiplantibacillus plantarum | WCFS1® | — not stated separately | Human RCT | ("WCFS1"[tiab] OR "NCIMB 8826"[tiab]) AND randomized controlled trial[pt] AND humans[mh] → 2 |
| Lacticaseibacillus paracasei | A234 | — not stated separately | No records | ("Lactobacillus paracasei A234"[tiab] OR "Lacticaseibacillus paracasei A234"[tiab] OR "paracasei A234"[tiab]) → 0; second spelling "A234"[tiab] AND (probiotic[tiab] OR Lactobacillus[tiab] OR Lacticaseibacillus[tiab]) → 0 |
| Bifidobacterium lactis (correct name: Bifidobacterium animalis subsp. lactis) | A026 | — not stated separately | No records | ("Bifidobacterium lactis A026"[tiab] OR "Bifidobacterium animalis A026"[tiab] OR "lactis A026"[tiab]) → 0; second spelling "A026"[tiab] AND (probiotic[tiab] OR Bifidobacterium[tiab]) → 1, opened and read, unrelated (PMID 35794494, porcine rotavirus in IPEC-J2 cells) |
| Lactobacillus acidophilus | A118 | — not stated separately | No records | ("Lactobacillus acidophilus A118"[tiab] OR "acidophilus A118"[tiab]) → 0; second spelling "A118"[tiab] AND (probiotic[tiab] OR Lactobacillus[tiab]) → 3, all opened and read, all unrelated (PMIDs 10998330, 12441069, 18430250 — "A118" is a locus and accession token there) |
| Lacticaseibacillus rhamnosus | A119 | — not stated separately | No records | ("Lactobacillus rhamnosus A119"[tiab] OR "Lacticaseibacillus rhamnosus A119"[tiab] OR "rhamnosus A119"[tiab]) → 0; second spelling "A119"[tiab] AND (probiotic[tiab] OR Lactobacillus[tiab] OR Lacticaseibacillus[tiab]) → 0 |
What of that is verifiable
A disclosure is verifiable when a reader can open a record that neither we nor the brand controls and see the same thing. Below is every dated artefact this page rests on. What is absent from this list is not verified, however reasonable it sounds.
- Photograph of the printed panel — 9_align_high_cfu_us_en_amazon_econtent_si_2500x2500_8_129433020.png captured The brand's own Supplement Facts image, served from images.ctfassets.net and transcribed by direct reading. Single variant, 30 capsules, GTIN 30772198681. This is the only transcription of this label that exists: no DSLD record was found under five query spellings, so nothing independent corroborates it.
- Photograph of the printed panel — 4_align_high_cfu_us_en_amazon_econtent_si_2500x2500_2_132766732.png captured The marketing image carrying the basis disclosure — "70B CFU^, including BIFIDO-35624™ + a blend of 5 strains" with "^AT TIME OF MANUFACTURE" set in roughly 6-point caps at the bottom edge of a 2000×2000 px asset. Recorded because the disclosure is real and its typography is the finding.
How many of its strains have a human trial for what it is sold for
Across the frozen founding cohort, 0 of 294 organism slots (0.0%) carry an organism with a human randomised controlled trial for any male-specific outcome. Not for this product — for all 27 of the founding teardowns, taken together. That is the number to hold against any men’s probiotic, including this one.
Two further founding-cohort figures for context: only 43 of 294 organism slots carry a strain with any human RCT at all under the exact code printed on the label, and only 2 of 27 products have a published human trial of the finished formula — the thing actually in the bottle. All three figures divide by the frozen cohort of 27 teardowns and 294 organism slots, never by every teardown on this site.
For this product specifically, 2 organism slots carry a strain with a human RCT under the exact designator printed on the label.
A word on what a designator has to clear before it counts here. The organism must be named with the exact strain code that appears in the trial, the trial must be in humans, and the code must return that trial when a reader searches it. A species name does not qualify: the trials in this field are of strains, and a species contains thousands of them.
What it costs per day
What certification it holds, checked at the issuer’s own register
A badge on a bottle is a claim like any other. Every row below was checked — or not checked — at the certifier’s own public register, never at the badge. Across the frozen founding cohort, 2 of 27 products hold a certification we could confirm at the issuer — a founding-cohort figure, not a count over every teardown on this site.
How it differs from the same brand’s women’s version
Across the fourteen same-brand men’s-versus-women’s label diffs run inside the frozen founding cohort, zero men’s or default formulas contain a probiotic strain, identified by designator, that was selected on male-specific evidence. In eight of the fourteen the sex-specific differentiator is a botanical or a mineral, and in five of the fourteen the women’s SKU is better-specified, cheaper, or both — “better-specified” meaning the panel that discloses more, never the panel with more in it. The rule behind each of those counts, and all fourteen pairs classified against it , are published in full. Diffs run on teardowns added since sit outside this fourteen, and outside the sentence above.
Compared against Align Probiotic Women's Dual Action. Diffed .
What the two panels differ by
- The women's SKU carries an at-expiry guarantee and no at-manufacture number at all — "† Provides 6 billion (6 x 10⁹ CFU) of probiotic cells until the 'best by' date." This SKU leads with 70 billion at manufacture and guarantees 15 billion, a figure that appears in no marketing copy anywhere. Panel 8_00030772032565_Label_Facts_111295726.png versus panel 9_align_high_cfu_us_en_amazon_econtent_si_2500x2500_8_129433020.png, both read 2026-07-31
- Every organism in the women's Feminine Health Probiotic Blend has a literature at its exact designator — La-14® and HN001™, 49 human RCTs between them — plus a botanical with its own. On this panel four of six designators return zero records at their exact code. PubMed ("La-14"[tiab] OR "La14"[tiab]) → 102 and ("HN001"[tiab] OR "HN-001"[tiab] OR "DR20"[tiab]) → 170, of which 49 RCTs, against the four zero-return queries on A234, A026, A118 and A119, all run 2026-07-31
- The women's panel carries no titanium dioxide. This one lists titanium dioxide (color) alongside maltodextrin, trehalose, sucrose and silicon dioxide. Both panels, transcribed 2026-07-31
What is identical across the pair
- Both carry Bifidobacterium longum subsp. longum 35624™ at 10 mg.
- Both are one capsule daily and both state their potency basis on the panel.
Every finding on this page, with the record behind it
Finding 1 four-of-six-designators-return-zero-literature
A234, A026, A118 and A119 return nothing at their exact designators. Every zero was re-queried with a broader second spelling, and every broad hit was opened and read: PMIDs 10998330, 12441069, 18430250 and 35794494, none of which concerns an Align strain — "A026" and "A118" are locus and accession tokens there.
The record: PubMed E-utilities, eight query strings published in full, run 2026-07-31
Finding 2 the-322-record-strain-has-two-rcts-and-neither-is-on-a-claimed-outcome
WCFS1 returns 322 records and is the L. plantarum genome reference strain, so the literature is genomics rather than clinical. Filtered to human RCTs it is 2 — PMID 20224007 and PMID 28045137 — both mechanistic mucosal studies in healthy volunteers measuring tight-junction protein localisation and mucosal gene transcription. Neither measures bloating, gas or abdominal discomfort. A raw count of 322 attached to "clinically studied" would be indefensible; the number is real and the inference would be false.
The record: PubMed ("WCFS1"[tiab] OR "WCFS-1"[tiab] OR "NCIMB 8826"[tiab]) → 322, AND randomized controlled trial[pt] AND humans[mh] → 2; PMIDs 20224007 and 28045137 opened and read, 2026-07-31
Finding 3 the-backed-by-science-strain-is-a-small-fraction-of-the-headline-cfu
Every research claim on the page is about 35624 — "BACKED BY SCIENCE: Align 70 Billion contains the unique Probiotic 35624™, which was developed by gastroenterologists and backed by 20+ years of scientific research." The panel gives 35624 at 10 mg, the identical quantity the brand's own current 24/7 Digestive Support label declares as 1 x 10⁹ CFU at manufacture. On the brand's own arithmetic that is roughly 1.4% of the 70 billion, with the remaining ~69 billion coming from a five-organism blend four of whose designators return nothing.
The record: Panel 9_align_high_cfu_us_en_amazon_econtent_si_2500x2500_8_129433020.png against panel 9_00037000143437_Label_Facts_21146132.jpg (Align 24/7 Digestive Support), both 2026-07-31
Finding 4 the-guaranteed-figure-appears-in-no-marketing-copy-at-all
21.4% of the headline is guaranteed to the best-by date — far better than the 1% on the two legacy Align SKUs, and the marketing image does disclose the basis: "70B CFU^, including BIFIDO-35624™ + a blend of 5 strains" with "^AT TIME OF MANUFACTURE" printed beneath. The debit is typography and omission. In a 2000×2000 px image "OUR LARGEST PROBIOTIC COMPLEX EVER" and "70B CFU" are set in display type while "AT TIME OF MANUFACTURE" runs in roughly 6-point caps at the bottom edge, and the 15 billion best-by figure exists only inside the Supplement Facts image. The number a buyer is actually guaranteed is the one number the brand never puts in prose.
The record: Panel 9_align_high_cfu_us_en_amazon_econtent_si_2500x2500_8_129433020.png and marketing image 4_align_high_cfu_us_en_amazon_econtent_si_2500x2500_2_132766732.png, both read 2026-07-31
Finding 5 a-proprietary-blend-of-five-organisms-at-108-mg
108 mg covers five organisms with no per-organism CFU and no per-organism mass, under a single dagger shared with the 35624 row. Any one of the five could be a token inclusion, and nothing on the label lets a reader find out.
The record: Panel 9_align_high_cfu_us_en_amazon_econtent_si_2500x2500_8_129433020.png, transcribed 2026-07-31
Finding 6 no-federal-record-exists-for-this-sku
Five query spellings returned nothing — the product name twice, the blend name, the strain name and the UPC in two formats — and the ten Align records surfaced by the blend query were opened individually and are all 35624-only legacy SKUs. Consistent with a recent launch, since DSLD entry lags market. It also means no independent transcription of this label exists: the panel is single-sourced on the brand's own image.
The record: NIH DSLD search-filter API, five query spellings, 2026-07-31
Finding 7 nomenclature-error-inside-a-freshly-typeset-panel
"Bifidobacterium lactis" is not a validly published standalone species; the correct name is Bifidobacterium animalis subsp. lactis. It sits on the same panel as three correct post-2020 genus names, so this is not legacy drift.
The record: Panel 9_align_high_cfu_us_en_amazon_econtent_si_2500x2500_8_129433020.png, 2026-07-31
Finding 8 amazon-keyword-syntax-shipped-on-the-brands-own-site
"WHAT'S A CFU? Colony Forming Unit measures the viable bacteria in a probiotic. Align 70 Billion CFU includes 6 different strains of probiotics for women and men." The strains are not for anybody in particular; "for women and men" is welded onto a definitional sentence where it is grammatically inert. Every gallery asset on this page is named align_high_cfu_us_en_amazon_econtent_… — the DTC page is Amazon A+ content, keyword string and all. The mechanism that routes ungendered products into men's roundups is visible here on the brand's own site rather than inferred from retailer titles.
The record: alignprobiotics.com 70 Billion page copy and gallery image filenames, read verbatim 2026-07-31 — open ithttps://www.alignprobiotics.com/en-us/products/probiotics-digestive-support/70-billion-cfu-probiotic-capsules — Address given, not linked: this site does not link out to the products it audits.
Finding 9 no-certification-of-any-kind
A word-boundary scan for NSF, USP Verified, Informed-Choice, Non-GMO Project, ConsumerLab, cGMP, "third-party", "certified by", Clean Label and BSCG returned zero hits on all ten tokens. Nothing is claimed, so there is no issuer register to check. Naive scans returned an NSF hit that resolves to the CSS property "tranSForm" — an artefact this project has now reproduced twice.
The record: alignprobiotics.com 70 Billion page, naive and word-boundary token scans, 2026-07-31
Finding 10 no-published-trial-of-this-six-strain-combination
("WCFS1"[tiab] AND "35624"[tiab]) returns 0. No study of the finished formula exists.
The record: PubMed E-utilities, run 2026-07-31
Finding 11 supplement-facts-published-only-as-a-raster-image
The panel exists only as a PNG, and the guaranteed 15 billion exists nowhere else — not in prose, not in the marketing, not in any federal record. The panel is being treated as a picture rather than as data, on a fully modern and fully fetchable site.
The record: Page HTML carries no text panel; panel image 9_align_high_cfu_us_en_amazon_econtent_si_2500x2500_8_129433020.png, 2026-07-31
What we could not verify
Every gap below is published rather than deleted. Most are derived automatically from the empty fields in this page’s own record, so a gap cannot be hidden by leaving a box blank.
- Third-party certification
- No certifier register has been queried for this product. An empty certification list on this page means not checked — it does not mean the product holds nothing.
- A US retail price, and therefore the cost per day
- The brand's own Next.js payload carries "price": null and the site has no cart. Retailer fallbacks all failed this session: Amazon geo-suppressed, iHerb 403, Walgreens HTTP/2 INTERNAL_ERROR, Target bot-blocked (HTTP 200, zero price tokens in the body). No aggregator figure was substituted, so no cost per day exists for this product.
- Per-organism CFU or mass for five of the six organisms
- 108 mg covers the Align Multi-Strain Probiotic Blend across five organisms with no per-organism CFU and no per-organism mass. Any one of the five could be a token inclusion, and the single footnote marker covering both panel rows means the 70 billion is a combined total rather than any organism's dose.
- Any independent transcription of this label
- No DSLD record exists for this SKU, confirmed with five query spellings — the product name twice, the blend name, the strain name, and the UPC in two formats. The ten Align records returned by the blend query were individually opened and are all 35624-only legacy SKUs. The panel here therefore rests on the brand's own label image alone.
- Any US retail price, and therefore any cost per day
- The brand payload carries "price": null with no cart, and every retailer fallback failed this session.
- Any third-party certification
- None is claimed. A word-boundary scan for all ten certification tokens returned zero hits, so there is no issuer register to query. The status is no certification, not unverified certification.
How to check every line of this page yourself
Nothing on this page requires trusting us. Every claim above resolves to a record you can open:
- Open the NIH Dietary Supplement Label Database record linked at the top of this page. It is the label the manufacturer filed, not our transcription of it.
- Compare the CFU line and the suggested-use line against the disclosure table above. If they disagree with us, we are wrong and we want to know.
- Take any strain designator from the panel and run the query printed beside it. The query is published precisely so you can get a different answer than we did.
- Divide the price by the number of days the label’s own dose instructions make the container last. That arithmetic is the whole of the cost-per-day row.
- Search the certifier’s own register for this product, not for the brand. A company can participate in a programme without this product being on its list.
Every organism on this panel has a designator. That puts it ahead of most of the category — and it is exactly why the label is worth reading closely, because four of the six designators lead nowhere.
A234, A026, A118, A119. Each was queried at its exact code, each returned zero, each zero was re-queried with a broader second spelling, and every broad hit was opened and read. None concerns an Align strain; in those papers “A026” and “A118” are locus and accession tokens. Four of six organisms on a 70-billion-CFU label have no literature at the code printed on the bottle.
The fifth is the trap. Lactiplantibacillus plantarum WCFS1® returns 322 records — the strongest-looking strain on the panel. It is not a clinical literature. WCFS1 is the genome reference strain for its species, the first L. plantarum ever sequenced, so it turns up across hundreds of genomics and comparative-microbiology papers. Filter to human randomised trials and 322 becomes 2, both mechanistic studies in healthy volunteers measuring tight-junction proteins and mucosal gene transcription. Neither measures bloating, gas or abdominal discomfort — the outcomes this product actually claims. Any roundup ranking strains by literature volume would rank WCFS1 above 35624 and be exactly wrong.
The 108 mg blend hides the split. Five organisms share one mass line with no per-organism CFU and no per-organism quantity, and a single dagger covers both the blend row and the 35624 row — so the 70 billion is a combined total, not any organism’s dose. All the research language on the page attaches to 35624, the one organism with its own declared mass and eight real RCTs. How much of the headline it actually contributes cannot be read off this label. The comparison against Align’s own 24/7 carton suggests a very small fraction; that comparison is an inference across two labels and is flagged as one at the top of this page.
Credit where it is due, twice. 21% of the headline is guaranteed to the best-by date — far better than the 1% on Align’s legacy SKUs — and the at-manufacture basis is disclosed rather than omitted. The debit is where: “AT TIME OF MANUFACTURE” runs in roughly 6-point caps under a display-type “70B CFU”, and the 15 billion figure a buyer is actually guaranteed appears in no prose anywhere — only inside the panel image. There is no federal record to check any of it against, and no third-party certification of any kind is claimed.