Catalog methodology
Purveyor Score
Purveyor Score is a metadata and listing-intelligence score for green coffee listings.
Purveyor Score makes green coffee metadata quality visible at the point of comparison. It rewards structured, comparable, buyer-useful listing information because buyers cannot evaluate what suppliers do not disclose.
The score is proprietary to Purveyors, deterministic, and intentionally narrow. It is not a cup quality score, supplier verification, certification, regulatory assurance, or a promise that a coffee is better than another coffee.
This methodology follows the argument in Who Profits When Coffee Data Stays Scarce?: the industry has made progress on price transparency, but product metadata remains inconsistent and relationship-gated. Purveyor Score turns that disclosure gap into an inspectable product signal.
What it measures
Purveyor Score is scored from the normalized coffee_catalog metadata already used by the catalog, API, CLI, and analytics surfaces. It favors fields that help a buyer compare similar listings across suppliers: provenance, process transparency, freshness, pricing comparability, and sensory context.
Confidence is separate from the score. Score measures metadata richness and buyer usefulness. Confidence measures how reliable and structured the inputs look, including recency, processing confidence, and evidence availability.
| Dimension | Max points | Examples |
|---|---|---|
| Provenance depth | 25 | Country, region, farm or producer notes, cultivar, grade, and appearance. |
| Process transparency | 25 | Base method, fermentation, additives, drying method, duration, disclosure level, and processing confidence. |
| Freshness and availability | 20 | Stock status, stocked date, arrival date, and last-updated date. |
| Pricing comparability | 15 | Per-pound price, tiered pricing, and wholesale classification. |
| Sensory context | 15 | Tasting notes, supplier cup score when present, roast recommendations, and useful catalog descriptions. |
Tier language
| Score | Tier | Meaning |
|---|---|---|
| 85-100 | Exceptional | Deep, structured metadata across most buyer-relevant dimensions. |
| 70-84 | Strong | Enough structured disclosure for confident comparison. |
| 50-69 | Developing | Useful listing, but key metadata is missing or less structured. |
| 1-49 | Limited | Sparse metadata. Treat as a starting point, not a complete sourcing picture. |
| 0 | Unscored | No usable score inputs are available. |
Why metadata matters
- Price answers whether a transaction may be fair. Metadata answers whether this is the right coffee to buy.
- Decision-critical fields such as arrival date, farm provenance, processing detail, cup score, and cultivar are less consistently disclosed than broad fields such as country.
- Normalizing listings across suppliers makes both disclosed data and missing data visible. That visibility creates an incentive for richer disclosure.
- The scraper and audit loop behind Purveyors treats data quality as something measurable: extraction gaps, field completeness, format validation, and source health all become signals the product can improve over time.
Not verification
Purveyor Score does not verify supplier claims, certify a coffee, replace buyer due diligence, or rate cup quality. It summarizes the listing intelligence Purveyors can currently see.
Implementation contract
- Purveyor Score v1 is stored on coffee_catalog as purveyor_score, purveyor_score_confidence, purveyor_score_tier, purveyor_score_factors, purveyor_score_version, and purveyor_score_updated_at.
- A database function computes the score from normalized fields and a trigger refreshes score fields when relevant metadata changes.
- The score can be recalculated when the formula changes. Consumers should read purveyor_score_version before comparing scores across major methodology updates.
- Raw supplier evidence remains withheld from public catalog responses unless a future product surface explicitly supports safe evidence inspection.
{
"purveyor_score": 82,
"purveyor_score_tier": "Strong",
"purveyor_score_confidence": 0.78,
"purveyor_score_version": "purveyor-score-v1"
}