Why most supplement advice is wrong
The supplement industry generates $50 billion per year in the US alone. Most of that revenue comes from marketing, not efficacy. Celebrity endorsements, influencer posts, and branded packaging dominate purchasing decisions — not peer-reviewed clinical data.
StackProtocol was built to answer a different question: what does the evidence actually say, ranked by research quality, not marketing budget?
How we score evidence strength
Every supplement on StackProtocol receives an Evidence Strength score (0–100%) based on four factors:
| Factor | Weight | What We Look For |
|---|---|---|
| RCT Quality | 40% | Randomized controlled trials with placebo groups. Double-blind preferred. Sample sizes over 100. Published in peer-reviewed journals (PubMed indexed). |
| Reproducibility | 25% | Has the finding been replicated by independent research groups? A single industry-funded trial scores low here even if the effect size is large. |
| Effect Size | 20% | Statistical significance alone isn't enough. We weight by whether the effect is clinically meaningful — not just p<0.05 but also Cohen's d > 0.4 or equivalent. |
| Safety Profile | 15% | Long-term safety data. Adverse event reporting. Drug interaction profile. Supplements with poorly characterized safety profiles are capped at 70% regardless of efficacy data. |
Evidence tiers explained
🟢 80–100% High Evidence — Multiple independent RCTs, large sample sizes, replicated effect sizes, strong safety profile. Examples: Creatine (95%), Vitamin D3 (90%), Omega-3/EPA (88%).
🟡 60–79% Good Evidence — Some RCT support but smaller studies, fewer independent replications, or emerging mechanistic data. Examples: NMN (65%), Rhodiola Rosea (72%), Berberine (78%).
🔴 Under 60% Emerging Evidence — Mechanistic rationale is sound, animal data is promising, but human clinical trial data is limited or conflicting. We include these only at our highest budget tiers and always with explicit caveats.
What we don't score
We deliberately exclude categories where evidence is consistently poor regardless of marketing claims:
- "Proprietary blend" products (can't evaluate undisclosed doses)
- Detox/cleanse products (no mechanistic basis)
- Weight loss "metabolism boosters" (no replicable effect beyond caffeine)
- Most collagen supplements for joint health (the peptides don't survive GI intact)
- Most multi-vitamins (bioavailability issues, poor form choices)
Our relationship with brands
StackProtocol earns affiliate commissions on purchases made through our links. We do not accept payment for rankings. Products are ranked by evidence first; affiliate availability is a secondary consideration for which brands we link to within a category.
In some categories (e.g., magnesium glycinate), we link to the lowest-cost Amazon option rather than a premium brand — because the evidence supports the compound, not a specific brand at a higher price.
"The goal is to be the site we wished existed when we started supplementing: honest rankings, real dosing, no asterisks."
Key references
Our evidence reviews draw primarily from: PubMed, Examine.com meta-analysis summaries, Cochrane Reviews, JAMA Network, and Nature Medicine. We do not cite manufacturer-funded white papers without independent corroboration.