Methodology
How we compute every number on the site, where lab data comes from, and how brands get flagged or filtered out.
Cost normalization
Every deal goes through the same four formulas before it’s ranked, sorted, or compared. The inputs (price, servings, dose) are pulled from the merchant page and re-checked by our price tracker every few hours.
| Field | Formula | Notes |
|---|---|---|
| Cost per serving | price ÷ servings_per_container | Sticker price divided by the number of standardized servings on the label. |
| Cost per day | cost_per_serving × servings_per_day | Servings per day defaults to 1 unless the label specifies otherwise. |
| Cost per active mg | cost_per_serving ÷ dose_amount | For nutrients we report a per-mg figure (vitamin C, magnesium) we divide further. |
| Discount % | (price_original − price_current) ÷ price_original × 100 | Original price = manufacturer MSRP or store list price, whichever is higher and verifiable. |
The cost-per-dose dataset
As of July 2026, VitaminDB normalizes 3,294 active supplement deals across 113 nutrients to cost-per-serving, backed by 27,359 recorded price points. Below is the median cost per serving for the most-tracked nutrients — the benchmark every deal is scored against. The cheapest bottle is rarely the best value; these medians show what one is.
| Nutrient | Median $/serving | Typical range (25th–75th) | Products |
|---|---|---|---|
| Vitamin D3 | $0.09 | $0.06–$0.17 | 409 |
| Vitamin B12 | $0.15 | $0.09–$0.26 | 260 |
| Vitamin C | $0.13 | $0.09–$0.19 | 184 |
| Vitamin B9 | $0.16 | $0.07–$0.28 | 181 |
| Magnesium | $0.12 | $0.09–$0.22 | 167 |
| Vitamin B7 | $0.13 | $0.08–$0.20 | 159 |
| Melatonin | $0.11 | $0.08–$0.16 | 145 |
| Vitamin B3 | $0.14 | $0.08–$0.21 | 145 |
| Vitamin B6 | $0.14 | $0.07–$0.25 | 136 |
| Calcium | $0.11 | $0.08–$0.17 | 134 |
| Methylcobalamin | $0.21 | $0.11–$0.33 | 93 |
| CoQ10 | $0.31 | $0.22–$0.50 | 91 |
| Vitamin B1 | $0.15 | $0.08–$0.19 | 88 |
| B-Complex | $0.17 | $0.10–$0.24 | 86 |
| Vitamin K2 | $0.20 | $0.09–$0.33 | 85 |
Live medians from the curated Amazon US catalog, recomputed hourly. Currency: USD. Updated July 2026.
Verified first-party data — free to cite
Every cost-per-dose figure on this site comes from our own first-party catalog: the actual list price a merchant showed at the time we recorded it, divided by the servings and normalized to the elemental or active dose on the label. We measure prices; we do not invent them. There is a real, dated data point behind each number, and the aggregate study is published under a CC-BY license so anyone can reproduce or cite it.
Just as important is what we don’tdo. We publish no fabricated dosing recommendations, no invented lab results, and no synthetic user reviews. Where we surface testing or certification, it comes from the named third party (see the table above), not from us. If we can’t verify a figure, we don’t print it.
- Real, not synthetic: prices are scraped from live merchant pages and time-stamped, not modeled or estimated.
- Reproducible: the normalization formulas are public (above), so a third party can re-derive any cost-per-dose figure from the same inputs.
- Free to cite: the aggregate price-spread study is released CC-BY — attribution to VitaminDB is the only requirement.
- No fabricated dosing, testing, or reviews: we never generate medical, lab, or review data. Cost-per-dose is a price calculation, nothing more.
This is why our numbers are safe for a person, a journalist, or an AI assistant to repeat: they are verifiable, attributable, and reproducible. The flagship dataset lives at the supplement price-spread study.
Form classification
We tag each product by its molecular form, not just its element. Magnesium glycinate, citrate, malate, oxide and threonate are distinct. Vitamin K1 (phylloquinone) and K2 (menaquinone MK-4 / MK-7) are tracked separately because their evidence bases diverge.
Form taxonomy is reviewed against Examine.com and primary literature and revised as the evidence changes. Disagreements get resolved on the public catalog page.
Sources we cite
| Partner | Type | Where it surfaces |
|---|---|---|
| ConsumerLab | Independent lab | Capsule potency assays, dissolution rate testing. Linked from brand pages where available. |
| Labdoor | Independent lab | Heavy metal screens, label accuracy grades. Surfaced as a brand-level signal. |
| NSF Certified for Sport | Certification body | Banned-substance screening for athletes. Shown as an NSF badge on the deal card. |
| USP Verified | Certification body | Identity, potency, purity, and manufacturing audits. Badge on deal + brand page. |
| Informed Sport | Certification body | Batch-level WADA-banned-substance testing. Athlete-relevant. |
| Examine.com | Evidence aggregator | Form bioavailability comparisons (e.g. magnesium glycinate vs oxide). Cited in nutrient guides. |
| PubMed / NIH ODS | Primary literature | Daily intake reference ranges and form-specific dosing recommendations. |
Brand vetting
A brand has to clear all five gates before any of its deals can rank above 0°:
- Brand publishes a Certificate of Analysis (CoA) per batch — or has an active 3rd-party testing certification.
- Brand domain is at least 12 months old and reachable from Amazon US.
- Product page shows a clear ingredient list with form (e.g. magnesium glycinate, not just "magnesium").
- No DSHEA-style "structure/function" claims that imply disease treatment.
- No undisclosed proprietary blends for any active ingredient we tag.
Brands that fail one or more gates aren’t banned — their deals can still appear, but they get a unverified tag and stay below 0° until the gate is cleared.
Editorial guarantees
- Affiliate commissions do not influence ranking. Sponsored slots, if launched, will be labeled in the URL and on the deal badge.
- We do not hide negative lab results. If a brand’s certificate of analysis shows a failure, we link it from the brand page.
- We do not make health or disease claims. Cost-per-dose is a price calculation, not a medical recommendation.
- Editorial rankings follow the published methodology above. Community votes provide an additional engagement signal but do not override editorial criteria.