US Customs Bill-of-Lading & Import Intelligence (ImportYeti)

Extract US maritime import intelligence for any company — its overseas supplier relationships, shipment volume, HS codes, TEU, multi-location addresses, and direct contacts — sourced from ImportYeti. The underlying data class is US CBP vessel-manifest (bill-of-lading) public record.

Built for equity/credit analysts (import volume as a COGS/revenue signal), B2B sales teams hunting a competitor's real suppliers, and supply-chain risk desks.

---

Why use this actor

• No account, no API key. Reads the public company page.
• Rich aggregate per company — not just a name: total shipments, 12-month shipments, average TEU/month, total shipping cost, top supplier, country mix, HS codes, time-series, phone, website, and up to five geocoded addresses.
• Stable extraction — pulls the structured company object directly from the page's embedded data (Next.js RSC flight), not brittle HTML scraping.

Input

Field	Description
companies	ImportYeti company slug (apple) or a full /company/<slug> URL.
maxConcurrency	Companies in parallel (default 3).
proxyConfiguration	Cloudflare-fronted; residential recommended at scale.

Output — COMPANY_PROFILE

Dataset envelope (_input, _source, _scrapedAt) + recordType: "COMPANY_PROFILE", then the ImportYeti company aggregate with upstream field names preserved:

{
  "_input": "apple",
  "_source": "S2-nextflight",
  "_scrapedAt": "2026-06-03T10:00:00Z",
  "recordType": "COMPANY_PROFILE",
  "company_slug": "apple",
  "profile_url": "https://www.importyeti.com/company/apple",
  "ld_name": "Apple - ... - company Report - Import Yeti",
  "ld_description": "Apple's top supplier is Hon Hai Precision Industrial ... with 2449 shipments",
  "company_name": "...",
  "company_address_country_name": "...",
  "total_shipments": 0,
  "shipments_12m": 0,
  "avg_teu_per_month": 0,
  "total_shipping_cost": 0,
  "company_website": "...",
  "company_phone_number": "...",
  "company_hs_codes": [],
  "company_time_series": [],
  "company_address_loc1": "...", "company_address_loc1_lat": 0, "company_address_loc1_long": 0
  // ... ~90 upstream fields total, verbatim
}

_source is S2-nextflight when the embedded aggregate is found, or S1-ldjson (summary only) as a degraded fallback. If neither layer yields data the record is an _error: "no_data_extracted" — never a silent empty object.

PRODUCT_BREAKDOWN (one per top HTS-code group embedded in the page — raw upstream fields):

{
  "_input": "apple",
  "_source": "S2-nextflight",
  "_scrapedAt": "2026-06-03T10:00:00Z",
  "recordType": "PRODUCT_BREAKDOWN",
  "company_slug": "apple",
  "hts_code": "8504.40.9999",
  "description": "Static converters",
  "shipments": 128,
  "most_recent_shipment": "26/07/2019",
  "bol_number": "FTNVHKOAHS001319",
  "product_description": "2100Pcs Of Power Adapter ... From China To United States Via Hong Kong By Truck"
}

How it works

1. Fetch GET /company/<slug> with a curl_cffi (chrome131) TLS profile (passes Cloudflare).
2. Parse the application/ld+json WebPage block for the summary.
3. Decode the self.__next_f RSC flight fragments and lift the company aggregate object with a balanced-brace walker anchored on avg_teu_per_month (with fallback anchors).
4. Emit one COMPANY_PROFILE per company.

Known limits & roadmap

• WAF: Cloudflare. Plain requests pass in testing; curl_cffi chrome131 used for safety, residential proxy at scale.
• Full per-shipment BoL rows are NOT available anonymously. Recon (2026-06-03) found the backend API https://data.importyeti.com/v1.0/company/<slug> returns HTTP 401 without a token — the SSR page is rendered server-side with that data, and the embedded JSON only exposes the company aggregate + the top-10 HTS product groups (each with a shipment count and one sample BoL number), which this actor extracts. A complete BoL feed would require ImportYeti's authenticated/paid API or a primary CBP/FOIA data source.
• RSC format drift: the flight chunk shape can change on a Next.js upgrade. The walker tries multiple anchor keys and fails loud (no_data_extracted) rather than emitting partial garbage — monitor the error rate.
• ToS / provenance: this reads ImportYeti's aggregation; the underlying CBP manifest is public record. For durable, high-volume use, evaluate a primary CBP/FOIA bulk feed.
• Verified live 2026-06-03: /company/apple → 200, 771KB, 90-field aggregate extracted cleanly.