Audit Methodology

How the audit works

What StackRoast checks, where the data comes from, and what it cannot do. Everything you need to decide whether the findings are worth trusting.

AI Transparency

Generated from structure, not guesswork

StackRoast does not produce free-form AI health advice. The audit engine applies a structured ruleset to a curated database of compounds, clinical dose ranges, known interaction patterns, and bioavailability data. Where evidence is weak, that uncertainty is labeled explicitly in the output using our four-tier system. You can see the tier on every finding.

136Curated ingredientsCurated, sourced, dose-ranged

346Sourced referencesPubMed, Examine, clinicaltrials.gov, and more

4-tierEvidence systemUncertainty labeled at every finding

Evidence Tiers

Not all evidence is equal

Every supplement finding in the roast carries an evidence tier. You see it, not just the verdict. Here is exactly what each tier means.

TierQualifies asIn your roast

TIER 1Replicated RCTs + Meta-analyses

Multiple randomized controlled trials, or a systematic meta-analysis, with consistent effect direction and dose specificity.

Highest-confidence findings. Specific dose ranges cited with study references. Verdicts stated directly.

TIER 2Single RCT or Strong Cohort Data

One well-designed randomized controlled trial, or multiple high-quality observational studies with dose specificity in humans.

Good confidence. Findings labeled Tier 2. Study-specific limitations noted where relevant.

TIER 3Mechanistic or Limited Human Data

Animal models with plausible human mechanism, or limited pilot trials in humans without dose confirmation.

Moderate uncertainty. Phrased as "limited evidence suggests." Effect direction included; dose ranges marked as extrapolated.

TIER 4Preclinical or Anecdotal

In-vitro or animal studies only. Community-sourced anecdotal patterns without clinical validation.

High uncertainty. Included for completeness; explicitly flagged as unvalidated in humans. No dose recommendation issued.

UNRATEDInsufficient Data

Compound not in our database, proprietary blends with undisclosed ratios, or sourcing insufficient for a verdict.

No verdict. The gap is flagged, the compound is listed, and we note what context you should seek from a clinician.

Full Scope

What gets checked

Dose vs. effective range

Every compound is measured against its clinical effective range. Underdosing (paying for nothing) and overdosing (diminishing returns or risk) are both flagged with specific numbers — not vague concerns.

Absorption conflicts

Known interaction patterns are checked pairwise across your entire stack: competitive absorption (zinc + copper, calcium + magnesium), receptor interference, and metabolic pathway overlap.

Timing conflicts

Time-sensitive compounds — fat-soluble vitamins, pre-workout stimulants, sleep aids — are evaluated for timing conflicts and optimal administration windows.

Redundancy analysis

Compounds whose mechanism is already covered by your diet or other supplements in your stack are flagged. Paying for amino acids you already eat is a common one.

Form and bioavailability

Supplement form changes how much you actually absorb. Magnesium oxide vs. glycinate. Creatine monohydrate vs. HCl. Folate vs. folic acid for MTHFR carriers. Form-specific bioavailability differences are checked.

Brand quality signals

Where you specify a brand, third-party testing certification status (NSF, Informed Sport, USP) and known quality signals are noted based on database coverage.

Limitations

What we do NOT do

Knowing the boundaries is part of using the tool correctly. These are not legal disclaimers. They are specific constraints that may affect your findings.

01

No bloodwork integration

The audit applies population-level clinical data, not your individual biomarkers. If you have blood panels showing specific deficiencies or excess, your physician can contextualize our recommendations against your actual numbers.

02

Cannot verify what is in the bottle

Label fraud and underdosing are documented problems in the supplement industry. We flag brands with known quality issues, but we cannot test your product. Third-party certification (NSF, Informed Sport) is the only real-world verification — we note which brands have it.

03

Individual genetic variation is not modeled

Polymorphisms in drug-metabolizing enzymes (CYP450 family, MTHFR, COMT, etc.) can significantly affect how you process certain compounds. Population-level dose ranges may not be optimal for your genotype. Genetic testing paired with a clinician is the only way to address this.

04

Interaction database is not exhaustive

We cover documented interaction patterns in our corpus. Novel compound combinations, rare interactions, and interactions with prescription medications are flagged as review items — not definitive verdicts. For prescription overlap, consult a pharmacist or physician directly.

05

Not a substitute for medical advice

StackRoast synthesizes what the research says. It does not replace clinical judgment. If you have a diagnosed condition, are pregnant, or take prescription medications, your physician should have final say on your supplement protocol.