How we score, publicly.
You can't buy your way to the top of NO1GEVITY. This page shows every lever that moves a product's score, every tier system we use, and every reason we accepted or rejected a given source. If a product ends up in a certain tier, it is because the specific rules on this page put it there.
Summary How does NO1GEVITY score longevity supplements? Show / hide ↓
The site scores products across six areas: dose match to human studies (50%), transparency (20%), value (10%), formulation (10%), safety (5%), and community ratings (5%). Dose is checked against randomised controlled trials, which compare treatment groups fairly, and each ingredient is weighted so one tiny ingredient cannot carry a whole blend. Transparency rewards public testing certificates and clearly listed amounts, while value considers the monthly price at the studied dose. Safety starts at 5 points, with bonuses for independent testing and recognised certifications, and deductions for active safety concerns. The site accepts peer-reviewed trials, trial registries, systematic reviews, and regulator documents as evidence, while rejecting unsupported marketing and testimonials. Research is repeated through several independent passes, but the site acknowledges that new studies, non-English research, and industry-linked reviews can still be missed.
What this means for you: For buyers, the score is meant to reward study-matched doses, openness, safety, and reasonable value rather than marketing alone. It is a transparent comparison system, not proof that every highly ranked product works.
moderate evidenceThe score in 30 seconds
A product's overall score is built from six weighted pillars. Every pillar is traceable to a primary source. The full formula, data sources, evidence tiers and source hierarchy are in Advanced.
Dose vs. studies — 50%
Does the product's stated dose match what human trials actually used? Checked against PubMed RCTs, not marketing. Now dose-weighted across ingredients with a 45% per-ingredient cap so no single trace ingredient can carry a blend.
Transparency — 20%
Public certificate of analysis, third-party testing, disclosed ingredient amounts. No CoA, capped score.
Value — 10%
Price per month at the trial dose, normalised across the category.
Formulation — 10%
Evidence-aligned delivery form. Phytosome curcumin, softgels for fat-soluble compounds and other proven-absorbing forms earn a bonus. Weight doubled in v2.2 because bioavailability decides how much of the labelled dose reaches the bloodstream.
Safety — 5%
Baseline 5.0. Third-party testing adds 2.0. Public lab certificate adds 1.5. NSF, USP or Informed Choice certification adds 1.5. Active safety signal subtracts 3.0. New in v2.2, positive safety now earns points, not just triggers caps.
Community — 5%
Aggregated consumer ratings, Bayesian-shrunk toward a 3.5-of-5 prior with 20 pseudo-reviews. Trustpilot, Labdoor and ConsumerLab full weight. Amazon 0.9, iHerb 0.8, Reddit 0.5, YouTube 0.4. Weight cut from 10% to 5% in v2.2.
1. How our research is actually done
Before we score anything, we run the same question through several independent research passes. This is the single biggest reason our numbers change over time, and the single biggest reason we hold ourselves to a higher bar than a marketing page.
Multiple independent runs
Every ingredient, product, and controversy runs through several independent research passes before we publish a number. Each pass starts from scratch. Where they agree, we publish. Where they disagree, we settle with primary sources first.
Read the full explanation
For every ingredient, product, and controversy on this site, we run the underlying research question through several independent research passes before we publish a number. Each pass starts from scratch: it looks for peer-reviewed trials, government registries, and named-critic responses without seeing what the previous pass found. Then we compare passes against each other. Where they agree, we publish. Where they disagree, we go back and settle the disagreement with primary sources before we publish anything at all.
This is why our resveratrol page mentions both the 137-RCT positive-endpoint hit rate and the 2025 GRADE meta-analysis showing no SIRT1 effect. One number is not the whole picture. If we only ran the question once, one of those facts would probably be missing.
What counts as evidence
Peer-reviewed RCTs, government registries, systematic reviews, regulatory documents. Not affiliate reviews, not marketing pages, not testimonials.
See the full breakdown
Peer-reviewed randomised controlled trials, government trial registries (ClinicalTrials.gov, EudraCT), systematic reviews and meta-analyses (especially GRADE-assessed ones), regulatory review documents (EFSA, FDA, EMA), and named-author critical responses in peer-reviewed venues.
Commercial comparison sites and affiliate-driven marketing pages. We look at these to understand what doses and combinations are actually being sold. We do not cite them as evidence for whether an ingredient works.
n=1 self-experimenters like Bryan Johnson's Project Blueprint. Their tracked biomarker data is real, and we cover it as a case-report with the sample size of one clearly on the badge. It is never treated as the same as a controlled trial.
Vendor blog posts, affiliate reviews without disclosed methodology, testimonial screenshots, and marketing citations that don't survive a click-through to the primary source.
Why we don't name research tools
The source is always the peer-reviewed paper, the registry entry, or the regulator document. The tool that surfaced it is not the source, and naming it would imply otherwise.
Read more
The point of a research pass is that it finds the same primary sources anyone else can find. If we told you "we used tool X to look this up," it would suggest tool X is the source. It isn't. The source is always the peer-reviewed paper, the registry entry, or the regulator document that the pass surfaced. That's what we cite. That's what you can go read yourself.
What breaks the process
Three failure modes we watch for openly: recency lag, non-English literature gaps, and industry-adjacent reviews that pass a first look.
See all three failure modes
A trial published last week may not show up in the first two or three passes. We rebuild ingredient pages when a new pooled analysis comes out, not on a fixed schedule.
Chinese-language RCTs and Japanese consumer testing sometimes carry evidence English-language passes miss. We flag these as open gaps rather than pretend they don't exist.
Some affiliate-driven sites publish content that looks like independent methodology. We check for affiliate links, coupon codes, and undisclosed brand relationships before we accept any source.
Where a claim on this site cannot be traced back to a primary source, that's a bug. Tell us.
2. Data sources we monitor
Every number on this site is traceable to a named, verifiable source. We do not run our own lab tests. Instead, we aggregate, cross-check, and score what independent parties have already published. We monitor 316 products across 127 ingredients, drawing from the broadest source base of any longevity supplement index we are aware of.
Scientific and regulatory databases Daily
NIH National Library of Medicine. Peer-reviewed RCTs, meta-analyses, and systematic reviews for every tracked ingredient. We pull new publications daily and diff against our existing study set of 730 studies.
New and updated clinical trial registrations, including trials that have changed status from recruiting to completed, which can shift an ingredient's evidence tier.
CFSAN Adverse Event Reporting System. Adverse event reports filed against specific supplement products and ingredients. A spike in reports triggers a safety review and a visible warning on the product page.
FDA Enforcement Reports. Class I, II, and III recalls of dietary supplements. Any recall affecting a product in our index is published on the product page within 24 hours.
European, UK, and Swedish regulatory databases Daily
Because many products in our index ship to European customers, we monitor EU, UK, and Nordic regulatory channels in addition to US agencies. These sources catch safety signals that US-only databases miss.
European Medicines Agency. EU-wide drug and supplement regulatory actions, safety referrals, and pharmacovigilance signals. We track EMA opinions on substances also used as longevity supplements.
Rapid Alert System for Food and Feed. EU-wide safety alerts for food supplements, including contaminated or mislabeled products. Alerts affecting products in our index are published within 24 hours.
European Food Safety Authority. Scientific opinions on supplement ingredient safety, tolerable upper intake levels, and health claim evaluations. EFSA opinions feed into ingredient evidence tiers and safety caps.
UK Medicines and Healthcare products Regulatory Agency. Post-Brexit supplement regulation, drug-safety updates, and Yellow Card adverse event reports for supplement products sold in the UK.
MHRA adverse event reporting system for medicines and supplements. UK-specific safety signals that may not appear in US databases. Spike in reports triggers a safety review.
Swedish Medical Products Agency. Nordic regulatory actions, supplement classification decisions, and safety withdrawals. Swedish-language regulatory alerts are translated and tracked for products sold in the Nordic market.
Swedish Food Agency. Supplement notification database, contaminant surveys, and dietary intake assessments. Swedish market-specific product registrations and safety communications.
Heads of Medicines Agencies. EU member-state regulatory network. Coordinates cross-border supplement safety actions and mutual recognition procedures across 30+ national agencies.
Independent product testing Weekly
Independent label-claim verification, contaminant screens, and identity testing. When ConsumerLab tests a product we cover, the result is cited on the product page with a direct link.
German independent product testing, comparable to ConsumerLab for the European market. Results cited with direct links where available.
Independent supplement quality testing with published assay data for purity and label accuracy.
International Fish Oil Standards (Nutrasource). Third-party certification of omega-3 product purity, oxidation (TOTOX/PV/AnV), heavy metals, PCBs, potency, and freshness.
Regulatory enforcement and deceptive marketing Weekly
Supplement-related enforcement actions. When a brand on our index receives a warning letter, we publish it on the affected product page with the letter's URL.
Deceptive advertising cases and consent decrees against supplement marketers. Published in our breaking-news feed when a covered brand is named.
Office of the Attorney General. 60-day notices for supplement products exceeding Proposition 65 chemical exposure limits.
Preprint servers Weekly
Preprints in biology and aging research, often appearing weeks or months before peer-reviewed publication. Included with a visible "preprint" label.
Preprints in health sciences, including clinical trial preprints for longevity interventions. Included with the same preprint caveat.
Stem cell and regenerative medicine Weekly
Because unproven stem cell clinics are an active consumer safety risk in the longevity space, we monitor a dedicated set of sources that no other supplement comparison site tracks.
Human Pluripotent Stem Cell Registry. Clinical study registrations for stem cell and iPSC-based interventions.
Independent tracking of cell therapy clinical trials worldwide.
California Institute for Regenerative Medicine. Funding decisions, trial updates, and regulatory news for stem cell research.
Paul Knoepfler lab. Independent commentary on stem cell clinic enforcement, adverse events, and regulatory gaps.
Official list of FDA-approved cell and gene therapy products. New approvals are published in our research feed.
Policy research on harms from unapproved stem cell interventions and FDA enforcement gaps.
Consumer review and sentiment platforms Per product
Consumer signals are used as operational signal only (delivery, service, refunds). They are never used as evidence of clinical effect.
Verified buyer reviews, brand-level trust scores, and complaint themes. 291 products covered.
Community discussion threads on supplement brands and ingredients. 194 products covered.
Credentialed reviewer videos (registered dietitians, physicians) without affiliate exposure. 20 products covered.
Verified-purchase review data, star ratings, and review volume for products sold on Amazon.
Health-focused marketplace review data for products sold on iHerb.
Independent evidence aggregator for supplement ingredients. Used as a cross-reference, not as primary evidence. 27 ingredients cross-referenced.
Research institutes and n=1 monitoring Per publication
National Institute on Aging. Multicenter preclinical testing of candidate longevity compounds (rapamycin, acarbose, 17a-estradiol, etc.). Results feed directly into ingredient evidence tiers.
n=1 self-experimentation protocol with instrumented biomarker tracking. Covered as a case report (Tier 4), never as controlled evidence. Protocol changes and dropped interventions are flagged.
Peer-reviewed publications from a dedicated aging research institute. New studies are incorporated into ingredient pages.
European aging research output, including studies on dietary restriction and cellular senescence.
Aging Research Center. Swedish epidemiological and clinical aging research, including cohort studies referenced on ingredient pages.
European Research Institute for the Biology of Ageing. Mechanistic and translational aging research from a leading European institute.
Industry and certification bodies Per product
Global Organization for EPA and DHA Omega-3s. Industry membership and quality standards for omega-3 products. GOED member status is cited on omega-3 product pages.
Marine Stewardship Council. Sustainability certification records for fish oil and krill oil products. Certification status and any objections are tracked per product.
We also monitor industry and certification bodies. For dietary supplements specifically, we monitor GOED (Global Organization for EPA and DHA Omega-3s) for omega-3 product quality standards and MSC (Marine Stewardship Council) for sustainability certification of fish oil and krill oil products. What we do not use as evidence: vendor-funded white papers without a peer-reviewed publication, affiliate review sites without a disclosed methodology, social media testimonials, and marketing materials that cite unpublished or non-peer-reviewed data. See the source hierarchy in section 5 below for the full Tier A through C and Rejected classification.
3. The overall score
Method changelog
v2.2 · September 1, 2026. Six changes. All 307 products were re-scored. The mean composite moved −0.11 points. The largest single move was 0.8 points. No product moved by more than 1.0 point.
- Aggregation fix. Multi-ingredient products now use a dose-weighted mean with a per-ingredient contribution cap. No single ingredient may supply more than 45% of the dose vs. studies score in a product with 3 or more ingredients. Above that limit the score blends toward the equal-weighted mean. This stops a trace-dosed high-grade ingredient from carrying a 12-ingredient blend.
- Longevity relevance axis. Every one of the 123 ingredients now carries a longevity_relevance rating: high, medium or low. High is direct human lifespan or healthspan evidence, such as an ITP result or a healthspan RCT, and keeps full weight (1.00×). Medium keeps 0.90×. Popular but longevity-unproven compounds keep 0.75×. Current split: 15 high, 78 medium, 30 low.
- Community rework. Bayesian shrinkage toward a 3.5-of-5 prior with 20 pseudo-reviews. A product with 12 reviews at 4.8 stars no longer outranks one with 8,000 reviews at 4.4. Platform weighting: Trustpilot, Labdoor and ConsumerLab at full weight, Amazon 0.9, iHerb 0.8, Reddit 0.5, YouTube 0.4. Weight in the composite cut from 10% to 5%.
- New Safety component, 5%. Positive safety is now scored, not only capped. Baseline 5.0. Third-party testing adds 2.0. A public certificate of analysis adds 1.5. An NSF, USP or Informed Choice certification adds 1.5. An ingredient under an active safety signal subtracts 3.0.
- Formulation weight raised from 5% to 10%. Bioavailability of the delivered form was under-weighted.
- Evidence-gated dose weight lowered from 55% to 50% to make room for the Safety component. Safety caps at over 200% of trial dose without an RCT and above the upper intake level are unchanged.
v2.1 · August 12, 2026. Following an independent methodology review we shipped four corrections plus three enhancements:
- Fix #4 — No more dose double-counting. Aggregation weight is now set by evidence grade (A=1.00, B=0.85, C=0.55, D=0.35, F=0.15), not by dose coverage. The dose penalty lives inside the ingredient score alone.
- Fix #10 — Value is catalogue-invariant. Fixed anchors at $0.10/day (10.0) and $5.00/day (3.0), linear in between. Adding a new product no longer shifts every other product's Value score.
- Fix #17 — Nomenclature. Ingredient pages now correctly say "out of 10", not "out of 100".
- Tier 4 / Tier 5 split. Tier 4 is human RCT at a different dose. Tier 5 covers mechanistic, animal, and marketing-only data. The badges no longer conflate the two.
- Enhancement: Safety signals. Product pages now display ingredient-specific safety signals (hepatotoxicity, upper intake limits, drug interactions) as visible warnings. Sources: NIH ODS, EFSA, FDA, PubMed. These signals do not modify the overall score — they provide context.
- Enhancement: Formulation multipliers. Ingredient evidence scores now adjust for bioavailability of the delivered form. Unformulated curcumin (under 1% absorption) receives a lower evidence contribution than phytosome curcumin. Ubiquinol CoQ10 scores higher than ubiquinone. Applied multiplicatively, defaults to 1.0 when no data exists.
- Enhancement: Trial-family dedup. Evidence grade upgrade checks now deduplicate publications from the same trial family (same NCT ID, same URL, or same first author + year + N). Prevents a single trial being counted as multiple independent studies.
All 293 products were re-scored on August 12, 2026. The top-5 ordering was unchanged. Twenty-three regression tests were added to the release pipeline to prevent regressions.
v2.0 · earlier 2026. Introduced the 5-pillar composite (Dose × Evidence gate, Transparency, Value, Community, Formulation) with safety caps applied after weighting.
If you only read one paragraph, read this one. We score every product on six things: whether its ingredients are proven (evidence), whether it contains enough of them (dose), whether the company shows its work (transparency), what it costs per effective day (value), whether the delivered form absorbs (formulation), whether the batch is tested and documented (safety), and what buyers report (community). Evidence comes first and acts as a ceiling: a product cannot score high by stuffing in large amounts of unproven ingredients, and it cannot buy a better score at any price.
Below, every component is explained twice: first in plain words, then as the exact rule we compute with. The tables are the model; there is no hidden logic behind them.
Three design principles behind the formula
The trial dose is the target
A product delivering 150% of the studied dose is not better than one delivering exactly the studied dose, it is just less studied. Our dose score peaks at the trial dose and tapers above it.
Evidence gates dose
A full dose of an unproven ingredient is not "half good". If the evidence is weak, no amount of milligrams can compensate: the evidence grade sets the ceiling, and the dose only decides how much of that ceiling the product reaches.
Potency is not mass
Ingredients are combined by how completely they cover their trial dose, not by milligram share. A gram of cheap filler cannot outvote 10 mg of a potent, well-studied compound.
Stress test: what if the weights were different?
We re-ranked the catalogue under several alternative weightings (dose vs. studies between 45% and 70%) and the rank order changed only marginally. The structure, the gate, the overdose taper, and the safety caps matter far more than the exact percentages. That is why we publish the structure in full.
Dose vs. studies: 50%
In plain words: this asks two questions about every ingredient. Is it proven? And does the product give you the amount that was actually studied? The proof sets the ceiling; the dose decides how much of that ceiling the product reaches. Too little scores low. Too much also scores lower, because nobody studied "too much", and more is not better, just less known.
Exact rule. For each active ingredient we compute a dose ratio r = ingredient mg ÷ trial dose mg, where the trial dose is the human-trial dose documented on that ingredient's page. The ratio maps to a dose score:
The multiplier runs from 0.52 (worst dose) to 1.0 (ideal dose), so evidence is always the ceiling. An A-grade ingredient at the trial dose scores 10. An F-grade ingredient at the trial dose scores 1.0, a full dose of something unproven stays a low score, by design.
How ingredients are combined into one product score
Each ingredient's weight combines its evidence grade with its dose adequacy (v2.2, September 2026). The grade weight is A = 1.00, B = 0.85, C = 0.55, D = 0.35, F = 0.15. The dose weight is the share of the trial dose the product delivers, capped at 1.0; below 10% of the trial dose it drops to 0.05. The two combine as w = grade weight × (0.3 + 0.7 × dose weight), so an ingredient never falls below 30% of its grade weight. The product-level score is Σ(w × ingredient score) ÷ Σ(w). Unknown ingredients enter at grade weight 0.20. Blends without listed amounts are capped at w = 0.25 for every hidden ingredient.
Contribution cap. In a product with 3 or more ingredients, no single ingredient may supply more than 45% of the weighted sum. Above that share the result blends toward the equal-weighted mean, in proportion to the overshoot. A 12-ingredient blend cannot ride on one well-dosed A-grade compound.
Longevity relevance. Each ingredient also carries a longevity_relevance rating that multiplies its score: high = 1.00, medium = 0.90, low = 0.75. High means direct human lifespan or healthspan evidence, for example an Interventions Testing Program result or a healthspan RCT. Low means popular but longevity-unproven, which can still be a strong ingredient for other outcomes. Across the 123 ingredients in the database: 15 high, 78 medium, 30 low.
History. v2.1 (August 2026) removed dose from the weight entirely, to fix a double-counting bug where an under-dosed ingredient was penalised twice. That went too far the other way: a 12-ingredient blend could hold three A-grade compounds at 2% of their trial dose and still average well. v2.2 (September 2026) puts dose back into the weight, but with a floor of 30% of the grade weight, so an under-dosed A ingredient is discounted without disappearing. A gram of D-grade filler still cannot outvote 10 mg of an A-grade compound.
Where do the trial doses come from?
Each ingredient's trial dose is an editorial judgment, not a fixed scientific constant. We set it by reading the published human trials for that ingredient and picking the dose that the best-available studies actually used on the endpoint that matters for longevity. The dose is paired with a range (the lowest and highest doses tested in published trials) and with direct links to the landmark studies on each ingredient page. You can check our work by clicking through to the papers.
These doses are not updated automatically. Our automated research sweeps (PubMed, ClinicalTrials.gov, FDA, CAERS) flag new studies and safety signals daily, and when a new trial materially changes what we know about a dose, the change goes through our editorial review before the trial dose moves. The old dose, the new dose, and the study that triggered the change are all recorded in the product's recommendation history with a direct link to the source.
If no human trial exists for an ingredient, we set research_dose_mg to 0 and the ingredient enters the score at a neutral weight (w = 0.1, dose score = 5). This means unproven ingredients cannot drag a product up or down by large amounts, they are present but quiet.
Transparency
Does the company show its work: exact doses, published lab certificates, named sourcing? You should not need to trust anyone's marketing, including ours.
Exact rule
Editorial 0–10, criteria published in full: we dock points for blends without listed amounts that hide per-ingredient doses, no publicly linked certificate of analysis (CoA), "tested by independent lab" claims with no document behind them, undisclosed sourcing, or missing lot-level testing. Each deduction is itemised on the product page.
Value
What does one effective day cost? A cheap bottle that under-doses is not cheap.
Exact rule
Inverse of price per day at studied dose (the cost of actually reaching the studied dose for one day), mapped to fixed anchors: $0.10 per day scores 10, $5.00 per day scores 3, linear in between, clamped at both ends. Sticker price and bottle size never enter directly.
The anchors are absolute (v2.1, August 2026), not catalogue-relative. Adding a new $500 product or removing the current cheapest one no longer shifts every other product's Value score. A product that scored 8.4 last week still scores 8.4 this week unless its own price per day at studied dose changed.
Community
What do buyers report about delivery, service, and daily use? Useful signal, but reviews can be farmed, so they can never outvote evidence.
Exact rule
Rating signals from external platforms, shrunk toward a prior of 3.5 out of 5 with a strength of 20 pseudo-reviews (v2.2, September 2026). A product with 12 reviews at 4.8 stars lands near 3.9 after shrinkage; a product with 8,000 reviews at 4.4 stars keeps almost its full rating. Platforms are weighted by how hard they are to game: Trustpilot, Labdoor and ConsumerLab at 1.0, Amazon 0.9, iHerb 0.8, Reddit 0.5, YouTube 0.4. Products with no traced signal default to 6.0.
The weight was cut from 10% to 5% in v2.2. Review counts are manipulable and the signal-to-noise ratio is low. The 5% weight is a ceiling, not a floor: community sentiment cannot on its own move a product past a better-evidenced one, and is never used as evidence of clinical effect.
Consumer signals section. The "Consumer signals" block shown on individual product pages is a separate, additive layer on top of the community score, not an input to it. It compiles verified review counts, star ratings, and named praise/complaint themes from public platforms (Trustpilot, Amazon, iHerb, Reddit, YouTube) with a direct link back to each source. As of this build it covers 309 products with 21,454,860 individual review or thread signals traced. Coverage is uneven by design, we only publish a card once we've located a genuine public source for that specific product, never a placeholder or an estimate.
Formulation
The same ingredient can be absorbed well or poorly depending on how it is delivered. A form proven to absorb better earns a small bonus; a form known to absorb poorly loses a little.
Exact rule
0–10, default 5. Above 5 only where peer-reviewed pharmacokinetic data shows the delivered form materially improves bioavailability for that specific ingredient (e.g. phytosome curcumin, softgels for fat-soluble compounds); 2–4 where the form is known to impair absorption; exactly 5 where form data is missing. The component can only differentiate on published data, it never punishes missing information.
The weight was raised from 5% to 10% in v2.2 (September 2026). Unformulated curcumin absorbs at under 1%. A phytosome version of the same milligram count reaches the bloodstream. That gap was worth more than 5%.
Safety
Is the batch tested, and can you see the paperwork? Until v2.2 safety only ever subtracted through caps. Now good documentation also earns points.
Exact rule
0–10, baseline 5.0. Third-party testing adds 2.0. A publicly linked certificate of analysis adds 1.5. An independent certification named on the label, NSF Certified, USP Verified, Informed Choice or Informed Sport, adds 1.5. An ingredient under an active safety signal subtracts 3.0. The result is clamped to 0–10. Maximum reachable score is 10.0.
Adverse-event report counts alone do not trigger the penalty. Report volume tracks how many people buy an ingredient, not how risky it is. Only an active recall or an active warning counts.
This component is separate from the safety caps below. The caps are a lid on the whole score. This component is 5% of the weighted sum.
Safety caps
A great average must never hide a dangerous number. Safety is not one ingredient in the recipe, it is a lid on the whole pot. Applied after the weighted sum.
Exact rule
- Any ingredient above 200% of the trial dose without trial support at that dose → composite capped at 7.0.
- Any ingredient above an established tolerable upper intake level (NIH UL, EFSA UL, or NOAEL-derived where no UL exists) → composite capped at 5.0, plus a visible warning on the product page.
The cap and its specific reason are always displayed on the affected product page.
Risk overlay
After the overall score is computed, a separate risk overlay checks each product against known Tolerable Upper Intake Levels (UL) and drug interaction data. If a product exceeds a UL or contains an ingredient with clinically significant interactions, a "Review advised" flag appears on the product page.
Exact rule
The overlay applies penalties on top of the overall score:
- Dose at or above UL: 1.0 point penalty. At 2x UL: 1.5. At 3x UL or more: 2.0.
- Known CYP3A4 interactions: 0.8 point penalty. Other drug interactions: 0.5.
- Dose above 500% of research dose: 0.5. Above 700%: 1.0.
UL values come from NIH Office of Dietary Supplements and EFSA. Interaction data comes from published pharmacology reviews. The overlay never increases a score. Click the flag on any product page to return here.
Evidence grades: A through F
Each ingredient is assigned an evidence grade based on the strength of human trial data:
Click any grade badge on a product page to return to this table.
A worked example
A product with two active ingredients, transparency 9.0, value 6.0, no formulation data (5.0), safety 8.5 (tested by independent lab with a public lab certificate), community 7.5. Both ingredients are medium longevity relevance (0.90×):
Score each ingredient
Combine into dose vs. studies
Dose vs. studies = (0.731×6.55 + 0.550×4.75) ÷ 1.281 = 5.78
The 45% contribution cap does not apply here. It only runs on products with 3 or more ingredients.
Apply the weighted composite
composite = 0.50×5.78 + 0.20×9.0 + 0.10×6.0 + 0.10×5.0 + 0.05×8.5 + 0.05×7.5 = 6.6
No cap applies (no ingredient exceeds 200% of trial dose or an upper intake level). Every product page exposes the same breakdown, so any score on this site can be recomputed by hand.
What this model does not capture, openly
No scoring model is complete, and we would rather list the gaps than let you discover them:
We do not run laboratory assays
Unlike physical-testing organisations, we do not chemically verify capsule contents. We rely on labels, published CoAs, and third-party test documents, which is exactly why Transparency carries 20% and undocumented claims score zero within it.
Evidence grades are editorial judgments
They follow written rules (trial design, replication, endpoint relevance, GRADE-assessed meta-analyses where available), but two qualified reviewers could reasonably disagree at the margins of adjacent grades.
Trial doses depend on the endpoint
Where trials used different doses for different outcomes, we document which endpoint sets the trial dose on the ingredient page.
The weights are judgment, not physics
No weighting scheme is objectively correct. Ours encodes one explicit editorial position, evidence first, safety as a veto, popularity last, and we publish the sensitivity result above so you can see how much (little) the exact percentages matter.
Formulation data is incomplete
Absent pharmacokinetic data the component defaults to neutral, so it currently differentiates only a minority of the catalogue.
4. Evidence tiers: Tier 0 through Tier 5
Every product carries an evidence-tier badge at the top of its page. This badge answers one specific question: what is the strongest published human evidence available for this exact product, in this exact formulation, at the dose sold? It is separate from the overall score and from the source-hierarchy tiers below. It is the single most important piece of transparency on the site.
A product's evidence tier is set by its strongest available source. So a Tier 3 product may still have Tier 5 marketing citations on its page, we list them, but the badge reflects the best evidence, not the worst.
5. External source hierarchy: Tier A, B, C, and Rejected
The evidence tiers above answer "what does the clinical literature say?". The source hierarchy below answers a different question: what does the outside world say about this product, and how much should we trust each voice? Every source we cite in "What others say" carries one of these labels.
Two-question test we run on every candidate source
- Does the source disclose its methodology? A named reviewer with a written process gets more weight than an anonymous star rating.
- Does the source earn money if you buy the product? If yes, we can still take facts (label doses, ingredient names) but never the source's rating.
Both answers are shown on the source card as a pill ("Affiliate: yes/no", "Paid placement: no"). The card also carries a "How we use it" line so you can see whether we're using it as a primary reference, a supporting cite, or background context.
6. Ads vs. editorial: the separation, in writing
Ranking = never bought
Brands cannot pay for placement, cannot pay to move up the composite, and cannot pay to influence any component score. This is our founding rule and we will state it in exactly these words for as long as the site exists.
Ad slots = clearly labelled
If we run ads, they will be visually distinct from ranked results, labelled "AD", and never appear inside the ranked list itself.
Affiliate links = disclosed
If we ever add affiliate revenue on outbound clicks, every such link will be marked as an affiliate link on the product page.
Editorial = named and dated
Every editorial article carries a byline and a publish/updated date. See the editorial hub.
See our advertising policy →, the three slots we sell, the things no advertiser can buy at any price, the editorial firewall, and the refund rule for bad-faith advertisers.
7. Every text written from scratch
We do not republish another site's product description, review, or scoring rubric. We synthesise from primary sources and rewrite in our own words. Where we quote a named expert, we quote them briefly and link the original. Where we cite a study, we cite the paper, not a secondary summary of the paper. If we ever fall short of that standard we will correct it publicly and note the correction on the affected page.
8. Corrections
If you find an error, a wrong dose, a mis-tiered source, a stale regulatory status, tell us and we will fix it visibly on the affected page and note the correction with a date. Silence on errors is how supplement pages lose the reader's trust; we would rather show our working.
9. Not medical advice
Nothing on this site is medical advice. Longevity is an emerging field. Regulatory status varies by market and by ingredient (see our regulatory tracker). Talk to a clinician who understands your situation before starting any supplement.