
For years, supplier risk assessments relied heavily on geography. Country-level risk scores linked to such factors as corruption, rule of law, political stability and labor practices gave procurement teams an easy way to screen thousands of suppliers, prioritize due diligence and apply enhanced controls without hiring an army of analysts. With a little additional configuration, that was good enough. Not anymore.
Geography is still a useful starting point, but the risk evaluation process can’t end there. A country score tells you where problems or risks have been documented, but it can’t tell you whether the specific supplier you’re about to sign has been affected. And in a post-COVID world, labor availability and challenges have shifted dramatically, with typically lower risk areas such as the U.K., Australia, U.S. and Canada experiencing a resurgence in underage and forced labor.
Regulators, customers and investors are now asking a sharper question than where a supplier operates: They want to know what material risks, or potential risks, have been documented for each supplier across broader considerations; how those risks were assessed, and what controls are in place to mitigate risk impact. Meeting these expectations requires moving beyond broad geographic classifications to evidence-based assessments. While this level of due diligence has traditionally been time-intensive, artificial intelligence is making this process more practical.
Country-level risk scores weren’t built for modern supply chains. Today, a single product might pass through multiple manufacturers, subcontractors and logistics providers before reaching its final destination. A supplier’s headquarters says little about where production occurs, how materials are sourced, or who is responsible for recruiting workers. Additionally, the prevalence of subcontracting within the category and lack of visibility beyond tier 1 can obscure significant material and reputational risks absorbed by the supply chain.
Relying exclusively on geography can create problems on both sides. Suppliers operating in countries with a history of labor or corruption problems face higher scrutiny, including more extensive questionnaires, additional documentation and on-site audits. Even well-run suppliers can end up buried in administrative burden simply because of their location, regardless of track record. It slows onboarding and pulls attention away from the suppliers that actually deserve a closer look.
Suppliers that use unethical labor practices also get overlooked because of their geographic location. The International Labor Organization estimates forced labor in the private economy generates $236 billion in illegal profits every year. Forced labor isn’t confined to the countries at the top of a risk index, and can show up in places a geography-first model would wave through.
Then there’s the data gap. In a recent survey of 500 U.S. procurement, supply chain, operations and environment, health and safety leaders, only 44% reported having fully centralized, real-time visibility into third-party workforce, safety and compliance risk. The rest are working from information that’s delayed, split across systems or pieced together by hand. With patchy visibility into supplier operations, organizations may fall back on a country code because it’s one of the few readily available data points.
None of this removes the need for due diligence, but it does make the case for looking beyond geographic risk assessments and static risk data. A better approach is to move from a geography-led model to a multi-factor one that evaluates suppliers based on defensible proxy indicators tied to business exposure and risk relevance, such as type of work performed, materials involved, and how critical the supplier is to operations. This model doesn’t ignore geographic risk trends; it simply stops assuming a supplier is risky (or safe) based on its country of origin.
Moving beyond country-based assumptions is also becoming increasingly important from a regulatory perspective. In China, Decrees 834 and 835 have heightened scrutiny around the use of country-based assessments and emphasize evaluating entities based on their own activities and characteristics, rather than broad national classifications. Regardless of jurisdiction, the broader trend is toward evidence-based, supplier-specific due diligence rather than decisions driven primarily by geography.
Much of the data needed to build a more complete supplier risk profile already exists — it’s just scattered across adverse media coverage, financial filings, legal proceedings, sanctions and watchlists, shipping records, and facility-level location data. Procurement teams managing thousands of suppliers don’t have time to manually parse through all of this information and identify which developments warrant attention, nor do they want to overburden suppliers with lengthy questionnaires and due diligence engagement.
AI can analyze large volumes of structured and unstructured data, identifying patterns and flagging potential risks that might otherwise go unnoticed. For example, it can scan news across multiple languages for more grassroots adverse media detection and signalers, flag labor disputes at a single facility, spot a supplier whose financial health is sliding, or match location-specific risks at the site level rather than the national border. Longstanding risk platforms and agencies can also use years of direct supplier data to better forecast, build risk profiles and perform trend analysis to give businesses refined predictive modeling that helps focus deeper review efforts. Instead of assigning suppliers the average risk associated with their country, organizations can build a more accurate risk profile based on what’s happening at a specific company and across their operations.
This creates three important opportunities for procurement and sustainability teams:
It points due diligence where it’s needed. If the evidence shows a supplier in a “high-risk” country is performing well, time can be better spent elsewhere. If a supplier in a “low-risk” country starts showing warning signs, you’ll know sooner.
It moves assessment from an annual event to an ongoing one. A questionnaire filled out once a year is a snapshot, and a lot can change in 12 months. Continuous monitoring means a new lawsuit, bankruptcy filing or news report about working conditions gets flagged when it happens, not at the next renewal.
It gives procurement and sustainability teams something they’ve been missing: a record of why a decision was made. When a customer, auditor or board member asks how a supplier was assessed following a lawsuit or media exposé on poor labor practices, admitting warning signs were ignored because “it was in a low-risk country” won’t hold up. A documented trail of the signals you reviewed and the actions you took will.
Better targeting, ongoing monitoring and a documented record of every decision add up to a much stronger risk program, one that directs attention where it matters and holds up when someone asks how you got there.
None of this works if the technology is treated merely as a shortcut. Following are a few things every team should keep in mind before handing supplier risk over to a model:
Data quality comes first. AI is only as reliable as the information feeding it. Outdated supplier records, duplicate entries and incomplete ownership data will produce confident-looking scores that point in the wrong direction. Before investing in new tools, take an honest look at the state of supplier master data and deploy “copilot” models — for example, AI as the foundation with human expertise reviewing its outputs — where possible.
Risk assessments must also be explainable. A risk score nobody can explain isn’t much better than a country label. Teams should be able to see which signals drive a rating so they can check the reasoning, correct it when it’s wrong, and defend it when they’re asked.
Human judgment has to stay in the loop. AI is very good at sorting through large volumes of information and surfacing patterns. It’s not as good at weighing context, understanding local conditions, or deciding what a fair remediation plan looks like. The strongest programs use AI to do the sorting, so that people can do the thinking and apply data where and when it makes more sense.
Despite growing interest in AI, adoption across supply chain operations remains limited. A recent McKinsey survey found only 19% of supply chain leaders saying they’re deploying AI tools at scale. For procurement and sustainability teams, this leaves room to rethink how they assess supplier risk before existing processes become even harder to manage.
Expectations around supplier due diligence will only continue to grow, but that doesn’t mean organizations need to abandon country-level risk scores or overhaul their entire programs. They can start by identifying where geographic assessments leave the biggest blind spots, improving supplier data quality and adding indicators that reflect actual risk exposure.
From there, AI-enabled monitoring can help teams move toward more continuous, supplier-specific assessments, with clear processes for validating findings and taking action where needed.
As expectations for supply chain accountability increase, organizations will need to show where risks are likely to occur, how they identified risks within their supplier network, and what they did in response.
Katie Martin is director of sustainability & innovation at Avetta.


















