How Yard Registry Works

Yard Registry is a structured intelligence pipeline — raw public listing data flows through ingestion, deduplication, entity resolution, enrichment, and scoring layers to produce a clean, scored, actionable business graph across Jamaica's 14 parishes.

01 Data Ingestion

Yard Registry currently ingests from 17 institutional Jamaican sources: 12 are named publicly as anchor sources on the landing page, and 5 additional regulators, registries and industry bodies also contribute records. Every source is checked against robots.txt before any fetch begins. Only six fields are ever stored per record: name, category, phone, address/parish, website URL, and source attribution. See the full source index for individual dossiers.

Public anchor sources (12 of 17, named on landing page)

• Companies Office of Jamaica — Public company registry. Full A–Z prefix sweep of the public company search interface. Active companies only by default.
• FindYello Jamaica — Jamaican business directory. Category-browsed via the dashboard or CLI. Respects rate limits with configurable delay (default 3s). Stops on 403/429.
• WorkandJam — Jamaican business directory. Sitemap-driven ingestion of public listings.
• OpenStreetMap — Geofabrik daily PBF snapshot for Jamaica (~38 MB). Named POIs with amenity, shop, tourism, or office tags extracted. No live crawling.
• Jamaica Tourist Board — visitjamaica.com licensed-property listings. JSON-LD structured data extracted.
• CaribbeanJobs — Active employer hiring intelligence. Used for workforce-velocity signals.
• Ministry of Health — Licensed health facility registry. Public government data.
• Ministry of Education — Public institution registry.
• Ministry of Labour & Social Security — Public registrations.
• Betting, Gaming & Lotteries Commission — Public licensed-operator registry.
• Jamaica Hotel & Tourist Association — Public member directory.
• Jamaica Stock Exchange — Listed-issuer public information.

Additional institutional sources (5 of 17, methodology-only)

• National Compliance Regulatory Authority — Public producer registries (food, baked goods, concrete).
• Mining Leases Registry — Public mining-lease holders.
• Jamaica Ports Authority — Public licensed operators.
• Jamaica Customs Agency — Public listings.
• Yellow Pages Jamaica — Legacy directory ingestion, superseded by FindYello.

02 Normalisation

Before any record is stored, the raw business name passes through a normalisation pipeline: lowercasing, punctuation removal, common suffix stripping (Ltd, Jamaica, Co., etc.), and whitespace collapsing. This produces a normalized_name field used by all downstream deduplication logic.

03 Entity Resolution & Deduplication

Raw listing records map to canonical business entities — the deduplicated, merged representation of a real-world business. Entity resolution uses two strategies:

• Phone match: identical normalised name + identical phone → high confidence duplicate.
• Location match: identical normalised name + same parish + same category (used only when no phone is available, to avoid false positives on common names).

Pairs scoring ≥ 0.92 similarity are auto-merged. Pairs scoring 0.70–0.92 are queued for human review. Each canonical entity receives a stable Yard Registry identifier of the form JBIP-{PARISH_CODE}-{SEQ:06d} (e.g. JBIP-KGN-000042).

04 Enrichment

After entity resolution, each entity goes through the enrichment pipeline. Three enrichers run:

• web_probe — Attempts to resolve the publicly listed URL. Classifies as live website, social-only, or unreachable.
• social_scan — Detects Facebook, Instagram, and other social profile URLs. Sets social_only=true when only social URLs are found.
• whatsapp_detector — Identifies WhatsApp contact links or phone numbers formatted as WhatsApp contacts.

Enrichment is batched in a background sweep running every 15 minutes. Force-re-enrichment is available via the CLI for entities where signals may have changed.

05 Digital Maturity Scoring

Each enriched entity receives a digital maturity tier based on its online presence profile:

06 Opportunity Scoring

The opportunity score (0–100) measures the readiness of a business for digital services. Higher scores indicate higher outreach value. Components:

Scores map to tiers:

• Tier 1 — Hot (≥75): Immediate outreach candidates
• Tier 2 — Warm (55–74): Strong opportunity with minor friction
• Tier 3 — Cool (35–54): Medium-term pipeline
• Tier 4 — Cold (0–34): Monitor only

Scores are recalculated daily at 06:00 JMT by the background scheduler.

07 Geo Intelligence

Businesses with lat/lon coordinates (from OSM, JTB, or manual entry) are indexed into H3 hexagonal cells at resolution 9, 7, and 5. Hex-cell aggregates (density, average score, website adoption rate) are recomputed weekly and exposed via the /geo API endpoints and the dashboard Map view.

08 Data accuracy disclaimer

Business data is aggregated from public sources and may contain errors, outdated listings, or duplicates that escaped the deduplication pipeline. Yard Registry data is intended as a starting point for market research and lead generation, not as a definitive source of record. We recommend verifying contact details before outreach.

Businesses wishing to be removed from the registry can email [email protected]. Removal requests are processed within 14 business days.