The NORM Framework: Measuring Claim–Evidence Alignment in Consumer Brand Marketing
A structured methodology for evaluating whether consumer-facing marketing claims are proportionate to evidence publicly presented or cited by the brand
NORM’s assessment principles, evidence boundaries, scoring areas, verdict meanings, and limitations are publicly documented. Exact scoring weights, thresholds, model prompts, anti-gaming controls, and implementation details remain proprietary to preserve system integrity and prevent manipulation.
Consumer brand marketing operates under conditions of significant information asymmetry. This paper publishes the draft NORM Framework v1.1, a proposed method for evaluating whether consumer-facing marketing claims are proportionate to the evidence brands publicly present or cite. It describes the proposed construct definition, evidence boundary rules, claim taxonomy, ten matching axes, six diagnostic areas, claim-level finding procedure, and provisional aggregation model.
NORM evaluates representation, not product reality. A future finding would reflect what a brand publicly discloses within the documented review scope, not what its products ultimately do. Public representations influence consumer and machine interpretation, so their evidentiary basis should be inspectable.
NORM is a claim accountability system. A record should improve only when a brand publishes stronger relevant evidence, makes sources traceable, narrows a claim to match the evidence, or removes a claim that exceeds its support.
1. Purpose
NORM v1.1 is designed to turn the distance between what a brand claims and what it publicly shows into an inspectable record. Aggregate scoring remains provisional until validation is complete.
The NORM Index is a 0–100 claim–evidence alignment index. Higher values indicate that a greater share of the brand’s important marketing claims aligns with evidence the brand publicly presents.
1.1 What NORM Measures
NORM compares consumer-facing marketing claims with evidence publicly presented or cited by the brand.
This is the complete scope. Every rule, score, and output derives from this comparison.
1.2 What NORM Does Not Measure
NORM does not determine:
- Whether a product works.
- Whether a product is safe.
- Whether a claim is legally compliant with FTC, FDA, or any regulatory standard.
- Whether the brand possesses private evidence it has not published.
- Whether the wider scientific literature ultimately supports or refutes a claim.
- Whether a consumer should purchase a product.
A high NORM Index means the brand’s public claims are well-supported by evidence the brand itself makes visible. It does not mean the product is effective, safe, or recommended.
2. The NORM Index
The draft NORM Index is a proposed 0–100 alignment index derived from claim-level assessments. Its interpretation and publication thresholds remain subject to validation.
The Index is not a percentage. It is not an efficacy score. It is not a safety rating. It is not a legal compliance finding.
2.1 Rating Bands
The Index is accompanied by a rating band that provides interpretive context. Band boundaries are criterion-referenced — each corresponds to a substantive level of claim–evidence alignment, not a percentile of the assessed brand population.
2.2 Category Ranking
Category ranking is a proposed output, not a current validated comparison. NORM will not publish v1.1 ranks until coverage, range, reviewer, arithmetic, and validation gates pass.
3. Six Score Areas
The v1.1 draft proposes six diagnostic areas. No public v1.1 assessment currently reports them.
Each area receives a diagnostic subscore. The overall NORM Index is not necessarily a simple average of the six areas. The Index is computed from claim-level assessments; the six areas provide a diagnostic decomposition.
4. Evidence Boundary
The evidence boundary is the single most important rule in the NORM framework.
4.1 Brand-Cited Sources
Evidence that the brand explicitly links, references by DOI or PMID, or identifiably cites on its consumer-facing pages.
4.2 NORM-Discovered Sources
Evidence found through NORM’s own research that is not linked or cited by the brand.
NORM-discovered sources may only authenticate a brand-provided citation, classify the cited paper, or assess its applicability.
5. Assessment Unit & Coverage
Under v1.1, NORM will assess a brand’s consumer-facing website as a whole. Each assessment will report pages assessed, capture date, methodology version, coverage, and confidence. v1.1 is under development and is not yet operational.
When NORM cannot capture sufficient content, the assessment is marked Incomplete. No score or rank is published.
6. Claim Categories
| Category | Example |
|---|---|
| Outcome / Efficacy | “Reduces cortisol by 40%” |
| Functional Benefit | “Supports gut health” |
| Mechanistic | “Works by activating AMPK pathway” |
| Composition / Dose | “Contains 600mg KSM-66 per serving” |
| Testing / Certification | “NSF Certified for Sport” |
| Superiority | “Most bioavailable form available” |
| Testimonial-Implied Outcome | A brand-published testimonial implying a health outcome |
| Puffery / Nonmaterial | “Premium quality” — excluded from scoring |
6.1 Materiality
Only material claims are scored. Puffery, navigation text, and extraction noise are excluded.
6.2 Testimonials
Brand-selected testimonials implying health outcomes are included. Purely subjective opinion and unfiltered third-party review feeds are excluded.
7. Claim-to-Evidence Matching
For each material claim, NORM assesses how well brand-cited evidence supports the claim across ten axes: product formulation match, population match, dose match, outcome match, duration match, study design appropriateness, reported result match, qualification match, risk of bias assessability, and language strength match.
Claim-to-Source Linkage
A source supports a claim when there is a direct citation, footnote, or hyperlink attached to the claim, or when the source appears on the same product or outcome page. A generic science page does not automatically support every claim.
8. Claim-Level Verdicts
9. Aggregation & Guardrails
A materially unsupported prominent claim cannot be hidden by averaging many well-supported claims. NORM applies a worst-material-claim guardrail. Its exact parameters are proprietary.
A future v1.1 assessment would publish total material claims assessed, claim-level findings, the assessed claims, cited sources, limitations, provenance, review status, and diagnostic areas.
10. Record Movement & Anti Gaming Controls
Under the draft method, brands could improve a future record by publishing stronger evidence, linking directly to relevant studies, aligning claim language with their evidence, or removing claims that exceed their evidence. NORM does not promise a fixed reassessment cadence.
11. Capture, Provenance & AI Processing
The v1.1 target system uses AI assisted capture with structured human review and versioned assessment rules. A v1.1 publication would require a versioned receipt. This target architecture is not a description of the internal calibration tests.
12. Validation Status
NORM v1.1 is a published draft under implementation and validation. Zero NORM v1.1 assessments have been published. Legacy v8 test runs are excluded from public coverage, scoring, ranking, and market statistics. NORM will not describe v1.1 as validated or calibrated until formal testing is complete and documented.
13. Limitations
- NORM assesses only what appears on a brand’s consumer-facing website at the time of capture.
- Claims in images and video are a known capture limitation.
- Small numerical score differences may not be meaningful.
- NORM does not assess private evidence, unpublished studies, or confidential regulatory filings.
- Low-confidence assessments are marked Incomplete and not ranked.
14. Corrections, Challenges & Appeals
Brands and readers may submit correction requests. A uniform challenge, adjudication, and public correction log is a v1.1 governance requirement under implementation.
15. Governance & Independence
NORM’s governance controls are under development. NORM does not describe itself as “structurally independent” or claim that “payment never influences scores” until auditable organizational controls substantiate those statements.
16. Version History
Legacy v8 — Internal Calibration
The legacy v8 pipeline used an automated six dimension composite model. Its test runs are retained internally for calibration and failure analysis. They are excluded from public coverage and are not comparable with future v1.1 findings.
v1.1 — Published Draft (Not Operational)
NORM v1.1 proposes claim-level assessment with ten matching axes, six diagnostic areas, and a structured finding procedure. It is under implementation and validation. No v1.1 assessment, score, or rank has been published.
Appendix A: Worked Example
Capture date: 2026-08-15 · Methodology: v1.1 (hypothetical — v1.1 is not yet operational)
Claim: “Clinically proven to reduce stress by 44%”
Location: Homepage hero banner · Type: Outcome/Efficacy · Materiality: High
Brand-Cited Source
A hyperlink to a published RCT (DOI: 10.xxxx/example) on the same product page. DOI resolves to a peer-reviewed journal article.
Ten-Axis Assessment
| Axis | Result | Reasoning |
|---|---|---|
| Product match | Partial | Key ingredient tested, not exact formulation |
| Population match | Partial | Adults 25–45 vs. “all adults” |
| Dose match | Direct | Study dose = product dose |
| Outcome match | Direct | Same perceived stress scale |
| Duration match | Direct | 60-day study matches use period |
| Study design | Appropriate | Double-blind placebo-controlled RCT |
| Result match | Partially supports | 29% reduction found, not 44% |
| Qualification match | Material omission | Subgroup-only effect omitted |
| Risk of bias | Low | Adequate randomization and blinding |
| Language match | Overstated | 44% claimed vs. 29% found, subgroup only |
Verdict: Partial Visible Alignment
The brand cites a relevant, well-designed study, but marketing language materially overstates the findings (44% vs. 29%) and omits a material limitation (subgroup effect).