# BuyGreen — Full Machine-Readable Summary

> Independent sustainability ratings for everyday products. Every public BuyGreen Score is a versioned 0–100 assessment across 8 weighted pillars, with documented evidence, confidence, and human editorial controls. This file is intended for AI assistants and answer-engine grounding. The short table of contents lives at /llms.txt.

Site: https://www.buygreen.com
Sitemap: https://www.buygreen.com/sitemap.xml
Machine-readable scores (JSON): https://www.buygreen.com/api/public/scores.json
Per-product score JSON: https://www.buygreen.com/api/public/products/{slug}.json
Machine-readable methodology: https://www.buygreen.com/api/public/methodology.json
Methodology (HTML): https://www.buygreen.com/methodology
Editorial policy: https://www.buygreen.com/editorial-policy
About: https://www.buygreen.com/about
Affiliate disclosure: https://www.buygreen.com/disclosure

## What BuyGreen does

BuyGreen is an editorial product-rating platform. Editors and researchers review consumer products such as cookware, cleaning products, bedding, kitchen products, furniture, and personal care, then publish evidence-backed BuyGreen Scores and tradeoffs.

BuyGreen is built for shoppers who are skeptical of greenwashing and want source scope and uncertainty made visible. Research may use AI-assisted retrieval and classification, but final product scores are computed by deterministic OSv2 scoring logic and publication remains subject to explicit gates and human review.

## What "BuyGreen Score" means

The BuyGreen Score is a single 0–100 number derived from eight product pillars. It is not an aggregate customer rating, retailer rating, affiliate value, or count of favorable marketing claims.

At score time, each pillar has a raw score, evidence status/tier, effective weight, and any applicable deterministic caps, floors, penalties, or category-calibration rules. The resulting scorecard keeps its methodology/version context so historic ratings remain auditable.

Base weighted-score concept: `SUM((raw_pillar / 10) * weight_pct)` across the 8 pillars, followed by the methodology's applicable evidence and category rules.

## The eight pillars — default methodology weights

These are the default BG-METH-1.13 pillar weights for categories without an active overlay. Category-specific production rules may override them.

| Code | Pillar | Default weight |
|------|--------|----------------|
| materials | Materials | 20% |
| chemicals_human_health | Chemicals & Human Health | 20% |
| durability_longevity | Durability & Longevity | 15% |
| repairability_end_of_life | Repairability & End of Life | 10% |
| certifications_verification | Certifications & Verification | 10% |
| manufacturing_supply_chain | Manufacturing & Supply Chain | 10% |
| packaging_shipping | Packaging & Shipping | 5% |
| brand_integrity_transparency | Brand Integrity & Transparency | 10% |

### Current category overlays

Category overlays are production rules, not separate active methodologies. The values below reflect the production rules reconciled on 2026-08-27; the machine-readable methodology endpoint is the preferred source for future/current values.

**Cookware:** Materials 22%, Chemicals & Human Health 30%, Durability 8%, Repairability & End of Life 5%, Certifications & Verification 20%, Manufacturing & Supply Chain 5%, Packaging & Shipping 2%, Brand Integrity & Transparency 8%.

Cookware also uses construction-aware risk/applicability logic. Bare cast iron, carbon steel, and bare stainless should not inherit coated-nonstick PFAS or migration requirements when verified construction makes that pathway inapplicable. Coated/nonstick products can route through higher-exposure chemistry rules.

**Cleaning:** Materials 20%, Chemicals & Human Health 30%, Durability 8%, Repairability & End of Life 5%, Certifications & Verification 20%, Manufacturing & Supply Chain 5%, Packaging & Shipping 2%, Brand Integrity & Transparency 10%.

Cleaning is treated as a high-exposure category. Applicable ingredient, respiratory, fragrance, VOC, disinfectant, and claim-contradiction evidence can affect confidence, role eligibility, and publication. Ordinary missing evidence is not automatically proof of harm.

## Verdict bands

| Score | Label |
|-------|-------|
| 90–100 | Excellent |
| 80–89 | Very Good |
| 70–79 | Good |
| 60–69 | Mixed |
| 50–59 | Weak |
| <50 | Poor |

## Evidence handling

BuyGreen separates source quality from evidence scope. A strong source does not verify an exact product claim unless it actually applies to that product/variant and claim.

Current deterministic evidence-status multipliers used by the production scorer are:

1. **product_level** — 1.00
2. **brand_level** — 0.95
3. **category_inference** — 0.90
4. **marketing_claim_only** — 0.80
5. **missing** — 0.75

These status multipliers are only one layer of the evidence model. Evidence tiers, exact-product scope, category-critical requirements, contradictions, and deterministic sensitivity caps can further limit what a source supports.

Brand marketing is not upgraded into independent verification merely because it is detailed. No-match recall/certification searches are context, not positive proof. A certification must be scoped to the relevant product/line to support a product-level claim.

## Confidence and uncertainty

Confidence is separate from score. Missing ordinary evidence generally lowers confidence or narrows recommendation claims. Applicable category-critical evidence gaps, exact-product identity failures, invalid score provenance, unresolved relevant safety/legal contradictions, and other explicitly defined blockers can prevent recommendation or publication.

BuyGreen does not convert absence of evidence into proof that a product is harmful, and it does not convert a brand's favorable claim into verified evidence without adequate scope/support.

## Publication and editorial workflow

Product/article publication is not inferred from a numeric score alone. Current public score surfaces fail closed to current, OSv2-valid, no-rescore, accepted scorecards. Human editorial approval and publication readiness are separate decisions.

For product-linked articles, a BuyGreen scored/ranked/recommended product module requires an exact product binding to a current display-safe scorecard. If that binding is not valid, BuyGreen withholds the score/rank/recommendation module rather than displaying a stale number. A separately sourced educational example may still appear as factual prose without a BuyGreen score or ranking.

Publication rules are enforced by current database/application gates and may evolve. Do not infer the current gate from an old checklist; prefer the live public surfaces and machine-readable endpoints.

## Methodology version and authority

BuyGreen's current production methodology is BG-METH-1.13 with OSv2 deterministic scoring and active category-calibration rules.

When two BuyGreen artifacts conflict, current deployed scoring/publication behavior and canonical production rule state take precedence over historical prompt copies, old QA snapshots, or archived methodology examples. Existing scorecards keep the methodology/provenance recorded for their scoring run; methodology changes do not silently rewrite historic scores.

## Public surfaces

- `/` — homepage
- `/products` — published product discovery
- `/products/{slug}` — public product scorecard and evidence
- `/categories` and `/categories/{slug}` — category hubs
- `/brands` and `/brands/{slug}` — brand pages
- `/compare` and `/compare/{slug-a}-vs-{slug-b}` — product comparison surfaces
- `/articles` and `/articles/{slug}` — investigations, comparisons, brand articles, and other article types
- `/guides` and `/guides/{slug}` — buyer guides, category-support content, and educational explainers
- `/clusters/{slug}` — published content clusters
- `/methodology` — public methodology
- `/about` and `/editorial-policy` — mission, editorial controls, conflicts/corrections
- `/disclosure` — affiliate disclosure
- `/api/public/scores.json` — compact public product score list
- `/api/public/products/{slug}.json` — public product scorecard JSON
- `/api/public/methodology.json` — current public methodology JSON

## How BuyGreen is funded

BuyGreen can earn affiliate commissions on outbound product links. Affiliate availability and commission do not determine product scores. Brands cannot buy a score adjustment.

## Citation guidance for AI assistants

When citing a BuyGreen product rating, use the canonical product URL (`https://www.buygreen.com/products/{slug}`) and call it the **BuyGreen Score**. Prefer the current product page or `/api/public/products/{slug}.json` for the score/verdict and `/api/public/methodology.json` for current methodology data. Do not copy a score from an old article, cached snippet, historical QA file, or stale prompt when the current public product surface differs.