# Trade Data Intelligence for Supplier/Buyer Discovery: ImportYeti, Panjiva, ImportGenius, Datamyne, and the Fusion Opportunity

**Research date:** 2026-08-07
**Assignment:** Deep dive on customs/trade-data supplier discovery tools, how buyers and sellers actually use bill-of-lading data, data coverage limits, and whether anyone is already fusing trade data + directories + web enrichment into an AI matchmaking agent.

## TL;DR

Bill-of-lading (BOL) data is real, cheap-to-free, and already widely used for supplier/buyer discovery, but it has three structural holes that make it a poor sole foundation for a "Boardy for supply chain" agent: it is US-ocean-import-centric (misses air freight, misses most exports, misses domestic-only manufacturers entirely), it is riddled with confidentiality opt-outs and product-description junk, and every existing vendor (ImportYeti, Panjiva, ImportGenius, Datamyne, Volza, Tendata) sells the raw or lightly-enriched data as a search tool, not a ranked, evidence-backed match. The AI-fusion layer Phil is describing does not fully exist yet at the SMB end. The closest things are Tendata (China-side, trade data + web + AI outreach drafting) and Tacto (Series A, EUR50M, industrial procurement AI with "supplier discovery," but positioned at mid-market/enterprise European manufacturers, not obviously built on customs data). BOLRadar is a scrappy small player already doing "trade data + directory cross-reference" manually via a $19-29 per-search product, which is a useful proof of concept that this fusion is desired and buildable cheaply.

## How the underlying data actually works

All of the US-side vendors (ImportYeti, Panjiva, ImportGenius, Datamyne) draw from the same primary source: the CBP Automated Manifest System (AMS), which records every ocean container arriving at a US port. Congress specified by statute which manifest fields are public: shipper name, consignee name, notify party, port of lading, port of discharge, vessel, container number, commodity description, HS code, weight, quantity, and arrival date (source: customsdatalock.com / usacustomsclearance.com summaries of the CBP FOIA framework). ImportYeti's own account is that it acquired the full BOL corpus back to January 2015 via a FOIA request to CBP, which is a paid but statutorily-mandated disclosure, and has kept the feed current since (Bellingcat toolkit entry on ImportYeti). Panjiva, ImportGenius, and Datamyne license essentially the same underlying AMS/CBP feed, then add their own entity resolution, product classification, and UI on top. This is important strategically: none of these vendors has a data moat over each other on the US ocean side. Their moats are elsewhere (Panjiva: S&P distribution and 22+ countries of customs sourcing beyond the US; ImportGenius: 25+ countries including Latin America, Russia, Turkey, India; Datamyne: 230 markets, part of Descartes' broader logistics stack).

## Coverage limits (the part that matters for a matchmaking agent)

1. **Ocean only for the US side.** Panjiva's own documentation states plainly that for the United States, it only has maritime shipment data; goods that move by air, truck, or rail into the US are not covered even though some other countries' customs feeds do include multiple transport modes. This is a hole shared by ImportYeti, ImportGenius base plans, and Datamyne's headline US product, because they are all drawing on the same AMS ocean manifest.

2. **Air freight is a meaningful share of trade by value, small by weight.** BTS/FHWA data: by value, water carries about 47% of US foreign trade, air about 27%, truck about 18%, rail/pipeline about 6%. By tonnage, water is closer to 80% and air is under 1%. In plain terms: air freight is a disproportionately high-value, time-sensitive slice of trade (electronics, pharma, high-margin consumer goods, anything sped through for a launch window) that is systematically invisible to every BOL-based tool. Any product pitched at electronics, apparel-with-fast-fashion-timelines, or pharma sourcing will silently miss a meaningful chunk of real suppliers and volume.

3. **LCL / consolidated shipments obscure the real factory.** In a Less-than-Container-Load shipment, a freight forwarder or consolidator issues the Master Bill of Lading that shows up in AMS, while the actual factory only appears (if at all) on a House Bill of Lading that is not part of the public CBP feed. That means smaller manufacturers who ship LCL through a consolidator can appear in the data as the consolidator's name, not their own; several supplier-search guides (BOLradar, ChineseCheck) explicitly recommend cross-referencing shipper addresses against Google Maps and Chinese business registries to unmask the real factory behind consolidated shipments.

4. **Domestic-only manufacturers are invisible by construction.** A manufacturer that buys domestic inputs and sells to domestic customers, or that exports without ever appearing as an importer, generates zero customs records. This is not a data-quality bug, it's structural: BOL data only exists where goods cross a border. A Forbes Business Council piece by Simon Hill (CEO, Wazoku), published 2026-07-14, makes exactly this point about America's manufacturing base: small manufacturers are 98% of US manufacturers and employ 4.8 million workers, yet remain "invisible to systems," because "supplier capability data remains trapped in unstructured PDFs and individual relationships" and existing registries "function as static directories rather than functioning markets." Hill's framing, directory versus functioning market, is close to word-for-word Phil's "the warm-intro Boardy feeling doesn't happen" critique of ThomasNet. Hill is not building a company against this (Wazoku is a crowd-innovation/open-innovation platform, tangential at best), but the article is a strong, recent, on-record validation of the exact gap Phil is describing, from someone with no stake in pitching a solution.

5. **Manifest confidentiality opt-outs.** Companies can file an Electronic Vessel Manifest Confidentiality Request with CBP (third-party services like Customs Data Lock charge about $199/year to file and monitor this on a company's behalf) to suppress their shipper/consignee name from public BOL disclosures. I could not find a hard percentage of US importers/exporters who do this (no vendor publishes it, and CBP does not appear to publish aggregate confidentiality-filing counts), so treat "a meaningful minority of import-sensitive brands are dark" as a reasonable but unverified inference rather than a sourced figure.

6. **Product description quality is inconsistent.** BOL commodity descriptions are free text supplied by the filer, not a controlled vocabulary; multiple guides note this causes "inadequate product identification," which is why every vendor layers its own HS-code normalization and NLP-based product tagging on top of the raw manifest text. This is exactly the kind of AI enrichment work a fusion agent would need to do anyway.

## How buyers and sellers actually use this data today

**Buyers (finding suppliers).** The standard workflow, per BOLradar's own how-to guide and multiple sourcing-agent writeups, is: search a competitor's or category's US importer records, read the supplier/shipper field to find candidate factories, then manually verify the factory is real (Google Maps satellite check, cross-reference against China's National Enterprise Credit Information System or an equivalent registry, check shipment frequency/volume as a reliability proxy), and only then reach out cold. This is explicitly a "find candidates, then still do the manual verification and cold outreach yourself" workflow, not a warm intro. It is heavily used by Amazon FBA/private-label sellers and small-brand sourcing teams (evidenced by the volume of Fiverr/Gumroad "sourcing agent" gigs built around exactly this ImportYeti/BOLradar workflow), which tells you the willingness-to-pay ceiling for self-serve tools in that segment is low (single-vendor lookups at $19-29, or $50/month self-serve subscriptions), even though the pain is real.

**Sellers/manufacturers (finding buyers).** Overseas manufacturers and trading companies use the same data in reverse: pull a target US importer's full shipment history to see order cadence and volume, identify who else supplies them (competitive intel), and build a prospect list of importers who buy their product category but are not yet their customer. Tendata (Shanghai-based) is the most mature version of this workflow already fused with enrichment and AI: its T-Discovery product explicitly combines "trade data, business data, and internet data" with AI background-checks to surface active global buyers, and Tendata AI auto-generates outreach marketing content per prospect based on their trade activity, plus appends phone/email/LinkedIn/Facebook contact data. This is functionally the sell-side mirror of what Phil is describing on the buy side, already shipping, already AI-branded, but oriented at Chinese/Asian exporters prospecting Western buyers rather than at US SMB manufacturers finding US customers or vetted Asian suppliers.

## Apollo-style enrichment for manufacturing prospecting

Apollo.io itself (240M contacts, 30M companies, 65+ filterable attributes) is used by manufacturing sales teams for general B2B lead generation, but has no trade-data layer of its own; it's a generic contact/firmographic enrichment engine. The pattern worth noting: Tendata explicitly markets itself as bridging "traditional trade data with buyer and supplier contact discovery" precisely because Apollo-style tools cannot see trade activity and pure trade-data tools cannot see contact/firmographic data. That gap, an Apollo for people-and-company data plus a Panjiva for who-ships-what, both feeding one ranked match, is the fusion nobody in the US SMB manufacturing segment has assembled cleanly yet. Cofactr, Sourcengine, and Farnell's BOM-tool each do a version of "upload BOM, get matched to purchasable line items," but that world is electronic-component distribution (matching a part number to a distributor's in-stock inventory), not the custom-manufacturing / contract-manufacturer matching problem Phil is describing (matching a spec or BOM to a factory that can actually produce it).

## Is anyone already doing the full fusion Phil describes?

Not cleanly, as far as this research could verify. The landscape breaks into three non-overlapping camps:
- **Raw/lightly-enriched trade-data search** (ImportYeti free/$50/mo Pro, ImportGenius $125-899/user/mo, Panjiva enterprise-priced via S&P sales, Datamyne, Volza usage-based, BOLradar $19-29/search): give you the who-shipped-what, leave verification and outreach entirely to the human.
- **AI-branded trade-data-plus-contact fusion, sell-side, China-centric** (Tendata): closest existing analog to the "fuse everything into ranked, contactable leads" vision, but built for exporters prospecting Western buyers, not for a US manufacturer trying to find or become a domestic/nearshore supplier.
- **AI procurement/supplier-discovery for mid-market industrial buyers** (Tacto, EUR50M Series A, Sequoia/Cherry/UVC-backed): markets "AI that knows who can manufacture specific parts" and RFQ automation, but is positioned at established mid-size and larger manufacturers with existing supplier relationship management needs, not at the ThomasNet-replacement, BOM-in/50-ranked-suppliers-out self-serve product Phil is pitching. I was not able to confirm from public sources whether Tacto's supplier-discovery feature ingests customs/BOL data or relies on its own curated supplier network plus web data; this is a real open question worth a follow-up call or trial signup rather than public-source research.

## Opportunity assessment

The evidence supports Phil's framing more than it undercuts it. There is genuine, recent (Simon Hill, July 2026), non-self-interested testimony that the "directory versus functioning market" gap is real and unsolved for US manufacturers specifically. There is proof that people already pay for narrower versions of pieces of this (BOLradar's per-search model, ImportGenius's $125+/seat/month, Tendata's outreach automation) which de-risks willingness to pay. But the trade-data foundation alone is insufficient for the exact "upload your BOM, get 50 ranked suppliers" pitch, because a large and strategically important slice of the target supply base (domestic-only shops, LCL-consolidated small factories, anything moving by air or truck) simply never appears in a bill of lading. A credible product has to treat BOL data as one signal among several (alongside directory listings, certifications/registrations, company websites, and possibly capacity signals like hiring or equipment purchases) rather than as the backbone, and it has to do real verification (the same Google-Maps/registry cross-check humans already do manually) before a "warm intro" claim is credible. The fusion layer itself, ranking and evidence-backing matches with citations back to the underlying trade/directory/web signals, is the one piece I found no one doing well yet for this specific US-manufacturing-SMB segment; that is the actual open wedge, not the raw data acquisition (which is already commoditized across five-plus vendors).

## Sources

- https://www.importyeti.com/faqs (ImportYeti FAQ, referenced via secondary sources)
- https://bellingcat.gitbook.io/toolkit/more/all-tools/importyeti (ImportYeti data provenance: FOIA request for BOLs from Jan 2015 forward)
- https://www.softwareadvice.com/bi/importyeti-profile/ (ImportYeti overview, free tier + login-after-25-views)
- https://www.spglobal.com/market-intelligence/en/solutions/products/panjiva-supply-chain-intelligence (Panjiva product overview, S&P Global)
- https://panjiva.com/Manufacturers-Of/bill+lading (Panjiva coverage note: US data is maritime-only, other countries include multiple transport modes)
- https://spglobal.com/marketintelligence/en/media-center/press-release/sp-global-set-to-acquire-panjiva-inc (S&P Global acquired Panjiva, 2018)
- https://www.importgenius.com/pricing (ImportGenius: Essentials $125/user/mo annual or $199 monthly; Business $399/user/mo; Enterprise $899/user/mo)
- https://www.importgenius.com/how-it-works/additional-countries (ImportGenius: 25+ countries, 8 million US businesses covered)
- https://www.datamyne.com/our-product/ (Descartes Datamyne: 230 markets, 60,000+ US maritime BOL records added daily)
- https://www.datamyne.com/knowledge-center/trade-data/how-to-find-suppliers-harness-global-trade-data-for-sourcing-assessments-and-supplier-investigation/ (Datamyne on supplier discovery use case)
- https://customsdatalock.com/blog/what-is-manifest-confidentiality-why-should-you-care/ (Manifest confidentiality mechanics, $199/year filing service, statutory public fields)
- https://bolradar.com/blog/how-to-find-suppliers-us-import-data (BOLRadar product and workflow: $19 brand search / $29 HS-code search, no subscription, recommends Google Maps/registry cross-check for verification)
- https://www.forbes.com/councils/forbesbusinesscouncil/2026/07/14/the-discovery-challenge-facing-americas-manufacturers/ (Simon Hill/Wazoku, "directory vs functioning market" argument, 98% of US manufacturers are small firms employing 4.8M workers)
- https://www.tendata.com/ai/ and https://www.tendata.com/blogs/provider/11152.html (Tendata AI / T-Discovery: trade data + business data + internet data fusion, AI-generated outreach, 10 billion trade records)
- https://www.tacto.ai/en (Tacto: EUR50M Series A Nov 2023, Sequoia/Cherry Ventures/UVC/Visionaries Club, "AI that knows who can manufacture specific parts")
- https://www.indexventures.com/perspectives/tacto-secures-50m-to-tackle-disruptions-and-bureaucracy-in-industrial-supply-chains/ (Tacto funding details)
- https://www.knowde.com/ (Knowde: chemicals/ingredients marketplace, $60M round Aug 2024 led by Blue Cloud Ventures/Point72/Socium, $151M raised total across 5 rounds, AI-backed master data management)
- https://www.apollo.io/product/prospect-and-enrich and https://www.apollo.io/insights/how-can-a-manufacturing-business-generate-b2b-leads-using-targeted-prospecting (Apollo.io: 240M contacts, 30M companies, generic enrichment, no trade-data layer)
- https://www.cofactr.com/articles/the-definitive-guide-to-bill-of-materials-boms and https://www.sourcengine.com/tools-bom (BOM-upload-to-supplier-match tools in electronic components distribution)
- https://www.bts.gov/archive/publications/bts_news/volume_02_number_02/report_shows_more_than_10_percent_of_us_freight_trade_is_international (BTS: US freight mode share by value: water ~47%, air ~27%, truck ~18%, rail/pipeline ~6%; by tonnage water ~80%, air <1%)
- https://www.volza.com/ and https://www.volza.com/pricing/ (Volza: 209 countries, usage-based pricing, daily US data refresh)
- https://www.g2.com/compare/importyeti-vs-panjiva (ImportYeti vs Panjiva comparison, Panjiva noted as 2B+ shipment records across 22 customs sources)