In short
Trade data is the record of what physically crossed a border: customs declarations, bills of lading, and the firm-to-firm transactions behind them. In 2027 it matters to sourcing teams for a reason it did not in 2023. Duties and import bans now depend on where goods come from and who made them, so the country-of-origin and supplier-network questions that trade data answers have become cost and compliance questions.
AI makes that data usable. Research published in 2025 and 2026 shows models that map suppliers beyond tier 2 from public sources, infer which inputs a product depends on from a billion cross-border transactions, and track how firms actually rebuilt their supply bases after the 2025 tariffs. This guide covers what changed, what the research says, which kinds of tools do what, and a five-step playbook for 2027.
What counts as trade data
Three layers matter for sourcing:
Customs and bill-of-lading records. Shipment-level data showing shipper, consignee, product description and route. US ocean import manifests are public, which is why tools such as Panjiva, ImportGenius and ImportYeti exist.
Firm-to-firm transaction networks. Aggregated, cross-border records that show who supplies whom. At scale they reveal dependencies no single company can see from its own ERP.
Origin and compliance signals. Where something was made and by whom, matched against sanctions, forced-labour and due-diligence lists.
The scale is now large enough to map whole industries. A 2025–2026 study by Neave O'Clery and colleagues used transaction data covering tens of millions of firms and around a billion cross-border transactions to infer the key inputs for more than 1,200 products, and found that European industrial nations and China dominate critical intermediate products such as metals, common components and tools (arXiv 2508.12315).
Why trade data matters more in 2027
US tariffs now key off origin and enforcement
On 20 February 2026 the US Supreme Court held in Learning Resources, Inc. v. Trump that "IEEPA does not authorize the President to impose tariffs" (opinion). The tariff regime that followed is tied to forced labour. Since 24 July 2026, USTR has applied Section 301 tariffs to 60 trading partners: 10% for economies that have committed to adopt and enforce a forced-labour import ban, 12.5% for those that have not (USTR fact sheet). Origin data now sets the duty rate.
Low-value shipments lost their shelter too. The US suspended duty-free de minimis treatment for all countries from 29 August 2025 (Federal Register).
Forced-labour enforcement is expanding
On 31 July 2026 DHS added 43 companies to the Uyghur Forced Labor Prevention Act Entity List, bringing it to 187 entities, which DHS called "the single largest-ever expansion of the List". The new entries are in aluminium, apparel, copper, cotton and tomatoes (DHS). A supplier that is clean today can sit one transaction away from a listed entity tomorrow, and only network data shows that.
EU deadlines have dates now
EU Deforestation Regulation: applies from 30 December 2026 for large and medium operators, and 30 June 2027 for micro and small ones (European Commission).
Corporate Sustainability Due Diligence Directive: the Omnibus I amendments published in February 2026 narrowed scope to companies with more than 5,000 employees and over 1.5 billion euros in turnover, applying from July 2029 (Covington). Fewer companies are directly in scope, but their suppliers will still be asked for the evidence.
Supply bases already moved
Laura Alfaro and Davin Chor track the shift in US imports: China's share peaked at roughly 21% in 2017, fell to around 13% by the end of 2024, and stood at approximately 9% by late 2025, "a level last seen when China joined the WTO in 2001" (CEPR VoxEU). Global trade kept growing regardless: the WTO now forecasts merchandise trade volume growth of 3.9% in 2026 and 4.1% in 2027 (WTO, 8 October 2026). WTO Director-General Ngozi Okonjo-Iweala: "The numbers reflect trade resilience in action."
What the research says AI can do with trade data
Map suppliers beyond tier 2
Most companies cannot see past their direct suppliers, and suppliers rarely volunteer their own supply base. Sara AlMahri, Liming Xu and Alexandra Brintrup at the University of Cambridge built a pipeline in which a large language model extracts supplier relationships from public sources into a knowledge graph. In an EV battery case it "extends visibility beyond tier-2 suppliers" and "reveals critical dependencies and alternative sourcing options" (International Journal of Production Research, 2025).
Show what firms actually do under tariffs
The most useful 2026 paper for sourcing teams comes from Boya Peng, Morris Cohen (Wharton), Arnd Huchzermeier (WHU) and colleagues, built on Panjiva import transactions. After the 2018 trade war, US firms diversified their supplier bases. After the 2025 tariffs they did the opposite: they cut active suppliers and consolidated within networks they had already built, faster than before and mainly through supplier exits. The shift towards North America seen in 2018 was largely absent the second time (SSRN, July 2026).
The lesson: in 2027 trade data is as much for pruning and requalifying an existing supply base as for finding new names.
Put language models where they help
A 2026 systematic review by Zhe Song and colleagues maps LLM capabilities onto the SCOR model and places sourcing under "ethical and sustainable sourcing" (IJPR, 2026, open access). An interview study at WU Vienna adds the practitioner limit. One manager put it plainly: "AI supports strategy creation, but final decision-making still requires human expertise and judgment" (Lakhal, Aazami & Kummer, Digital Business, 2025).
Where adoption really stands
The 2026 numbers show intent running ahead of practice:
The Hackett Group's 2026 Key Issues Study found 43% of procurement organisations actively pursuing AI deployment, nearly double the year before, but only 12% at large scale (Hackett Group).
Gartner found 63% of senior procurement leaders expect AI to significantly improve performance within three years, yet only 25% have redesigned jobs or roles as a result (Gartner, 8 October 2026).
"Successful AI implementation at the functional level is fundamentally a role redesign challenge," said Fareen Mehrzai, Senior Director Analyst in Gartner's Supply Chain practice. Amy Hillcox of The Hackett Group made the same point from the other side: "Procurement is moving beyond isolated digital improvements and beginning to confront what AI really changes: how work gets done."
Practitioners want answers rather than dashboards. "I want AI to tell me what I need to know," Nordstrom CPO Karoline Dygas told a Manifest 2026 panel (Supply Chain Dive).
What practitioners say
People who do this work are less interested in the tools than in the evidence they produce.
Origin is the first problem. "Semiconductor suppliers in particular have been very slow in providing country of origin information," Graham Scott, VP Global Procurement at Jabil, told the Art of Procurement podcast in 2025. Waiting for suppliers to volunteer it is not enough.
Customs specialists say the same about proof. Rennie Alston, director of trade compliance at CargoTrans, warned that customs treats you as "guilty on the appearance of noncompliance, and until you can demonstrate reasonable care" (The Loadstar, September 2026). If you switch supplier or origin to cut duty, you need a record of why.
On AI, buyers want help with volume, not a replacement for judgement. John Eustis, SVP of US group procurement at Toray Industries (America), said his team uses AI agents "to just take a lot of the grunt work, the easy stuff, off their plate, so they can focus on the more important stuff" (Supply Chain Dive).
Shipment data also has blind spots. Wharton research found that "about one in four maritime shipments have redacted identities" (Knowledge at Wharton). Use trade data and AI to build the shortlist and the evidence file, then let people check it.
Tool types for trade data in sourcing
Tools differ more by job than by brand. Match the category to the question you are asking.
Category | What it answers | Examples | Watch for |
|---|---|---|---|
Customs and bill-of-lading databases | Who ships what to whom, by volume and route | Panjiva, ImportGenius, ImportYeti, Descartes Datamyne | Strongest on US ocean imports; thinner for air, road and intra-EU trade |
Forced-labour and risk screening | Is this supplier, or its network, exposed to listed entities, sanctions or high-risk origins | Supplier-risk and sanctions screening platforms | Coverage of tier 2+, how often lists refresh |
AI supplier discovery and vetting | Who else can make this, and can I trust them | Speya: describe the need in plain language; AI agents search millions of supplier profiles across 250+ data sources and return AI-vetted shortlists with sources beside every name | Whether every claim comes with a source you can check |
Sourcing-event optimisation | Which bid mix is best once you have suppliers | Keelvar and similar | Needs a qualified supplier list first |
Most teams need more than one. A common 2027 stack: trade data to see current exposure, screening to flag risk, and AI discovery to find and vet alternatives.
A 2027 playbook
Map origin exposure. Match your current suppliers' production countries against the Section 301 rates and the UFLPA Entity List. Start with the categories DHS named: aluminium, apparel, copper, cotton.
Prune before you add. As Peng et al. found, firms that came through 2025 consolidated. Use shipment data to see which suppliers you actually use and which only exist on paper.
Find alternatives with evidence. Use AI discovery for lower-exposure origins, and insist that every capability, certification and ownership claim comes with a source.
Keep a human sign-off. The research and practitioners agree: AI builds the long list and the evidence file; people decide.
Redesign the role, not just the tool. Gartner's numbers show where projects stall. Decide who owns trade-data monitoring before you buy anything.
Events to watch in 2027
ProcureCon West, 7–10 March 2027, Las Vegas. Includes the keynote panel "Make Sure To Check Your Blind Spots: Exposing Risk Across Tier 2 & 3 Suppliers" (agenda).
Gartner Supply Chain Symposium/Xpo, 3–5 May 2027 Orlando and 24–26 May 2027 Barcelona (Gartner).
ISM World 2027, 16–18 May 2027, Washington DC (ISM).
Frequently Asked Questions
Is bill-of-lading data public?
US ocean import manifest data is publicly available, which is why commercial customs-data tools can resell it. Coverage is weaker for air freight, road freight and trade inside the EU, so a gap in the data is not proof that a shipment did not happen.
Can AI map tier-2 suppliers?
Partly. Research from the University of Cambridge shows language models extracting supplier relationships from public sources into a knowledge graph that extends visibility beyond tier 2. The result depends on what is publicly reported, so treat it as a lead to verify, not a complete map.
How do the 2026 US tariffs affect supplier selection?
Since 24 July 2026, Section 301 tariffs of 10% or 12.5% apply to 60 trading partners depending on their forced-labour import rules. A supplier's production country now affects landed cost directly, so origin data belongs in the sourcing decision.
When does the EU Deforestation Regulation apply?
From 30 December 2026 for large and medium operators, and from 30 June 2027 for micro and small operators.
Sources
Every source above is linked inline. Key references: Supreme Court, Learning Resources, Inc. v. Trump (2026); USTR Section 301 fact sheet (July 2026); DHS UFLPA Entity List release (31 July 2026); WTO Global Trade Outlook (8 October 2026); Gartner CPO survey release (8 October 2026); The Hackett Group 2026 Key Issues Study; Peng, Cohen, Huchzermeier et al., SSRN (2026); AlMahri, Xu & Brintrup, IJPR (2025); O'Clery et al., arXiv (2025–2026); Song et al., IJPR (2026); Lakhal, Aazami & Kummer, Digital Business (2025); Alfaro & Chor, CEPR VoxEU.