What Supply-Chain Data Shows That Import and Shipping Data Miss
September 14, 2026
Altsets
Research by Altsets Research
Shipping data measures physical merchandise flows. Supply-chain relationship data measures economic dependence between companies, including commercial relationships that are not fully observable from ports, commodity codes, shipping weight, or customs value.
Data used:Altsets Supply Chain Intelligence: 90k+ entities, 400k+ relationships, 20+ years of history.
Key findings
- U.S. Census trade products organize merchandise flows by commodity, country, port or district, value, quantity, shipping weight, and transportation method rather than by complete public-company economic dependency.
- The supplied 561M USD HPE-Microsoft relationship and HPE's public Microsoft alliance illustrate a commercial connection spanning hardware, Azure Local, licensing, software, support, and hybrid services that cannot be reduced to a merchandise shipment count.
Shipping and import data measure physical merchandise moving through trade channels. Supply-chain relationship data measures which companies depend economically on which other companies. The overlap can be useful, but the two datasets are not substitutes because many commercially important relationships involve software, licensing, services, equipment economics, domestic activity, or company-level dependencies that cannot be reconstructed from a port-level commodity flow.
Merchandise data starts with goods, countries, ports, and quantities
The U.S. Census Bureau's trade products are organized around merchandise. USA Trade Online provides imports and exports by Harmonized System or NAICS classification, country, customs district, port, value, quantity, shipping weight, and method of transportation. That makes trade data extremely useful for questions about commodity flows, country exposure, shipping volumes, and changes in physical imports or exports.
Those fields do not inherently identify the economic relationship between two public companies. A rise in imports of one product category can be important without revealing which listed customer ultimately depends on which listed supplier, how material the relationship is to either company, or whether the relationship includes activities that never cross a customs boundary.
HPE and Microsoft show why company relationships can extend beyond merchandise flows
The supplied Altsets data contains a 561M USD relationship between Hewlett Packard Enterprise and Microsoft. HPE's own Microsoft alliance materials describe Azure Local, Windows Server licensing, GreenLake, integrated support and services, hybrid cloud infrastructure, software delivery, and jointly delivered enterprise solutions.
Some of that commercial ecosystem is tied to physical HPE hardware, but the relationship cannot be understood as a container count moving from Microsoft to HPE or HPE to Microsoft. Microsoft licensing, Azure services, hybrid software, support, and integrated offerings are part of a business relationship whose economic relevance is broader than merchandise trade.
That is where a company relationship graph answers a different question. It tells the investor that HPE and Microsoft are economically connected and gives the relationship a measurable place in the research process. Trade data can then add geographic and physical-flow context where merchandise is actually involved.
Shipping data is often stronger on physical activity and weaker on complete company economics
If an investor wants to know whether imports of a specific battery chemistry are accelerating through U.S. ports, trade data may be the better starting point. If the investor wants to know which public supplier is economically dependent on a particular customer, company relationship data is closer to the question.
The distinction matters because physical volume and economic importance can diverge. A high-value piece of semiconductor equipment can matter enormously without creating a large number of shipments. Software and service relationships can matter without producing a merchandise record at all. Domestic production can bypass international customs data entirely. A company can also buy through intermediaries, subsidiaries, distributors, or contract structures that make the path from shipment to final economic dependency difficult to infer from trade records alone.
Combining the two can produce a stronger conclusion than either one alone
A useful workflow starts with the relationship graph to identify the companies and direction of dependence, then uses trade or shipping data when the thesis depends on actual physical flows. If a known supplier relationship appears economically important and the relevant import category is accelerating, the two datasets can reinforce one another. If shipping volume weakens while the commercial relationship remains important, the discrepancy can motivate questions about inventory, domestic sourcing, product mix, or a change in how the relationship is fulfilled.
The key is not to force one dataset to answer the other's job. A relationship estimate should not be treated as a shipment count. A customs record should not automatically be treated as proof that the named importer and exporter represent the final economically material supplier-customer pair.
The conclusion is that shipping data tracks movement while relationship data tracks dependence
Shipping and import data are strongest when the investment question is about what physical goods are moving, where they came from, where they entered, and in what quantity or value. Supply-chain relationship data is strongest when the question is which companies depend on one another and how economically important that dependence appears. Altsets' moat is not that trade data is unnecessary. It is that a company-level dependency network captures a different layer of the economy that physical merchandise records alone were never designed to provide.
The alternative-data comparison explains how relationship data differs from several other alternative datasets. The public-sources proprietary-data guide explains why normalizing fragmented evidence into one historical company graph adds value even when some underlying sources are public.
For relationship definitions and evidence limits, read the Altsets methodology.