Autonomous sourcing, predictive insights, and conversational co-pilots all promise a new era of efficiency for procurement. AI is rapidly becoming embedded in day-to-day work and adoption is accelerating fast. According to recent research, 94% of procurement executives now use generative AI at least once a week, representing a 44-point jump in adoption from the prior year.1
Yet despite this momentum, one reality keeps surfacing: Before procurement can scale AI, it must fix the supplier data foundation AI depends on. As Bertrand Maltaverne, Principal & Global Procurement Advisory Practice Leader at The Hackett Group, put it, “And at the end of the day, this is a concern that has been top of mind for years, and that has not really been addressed.”
How AI in Procurement Is Being Used Today and Why It Breaks
AI is no longer experimental in procurement. Teams are applying it across critical workflows, including:
- Spend analytics and classification to surface savings opportunities
- Supplier discovery and recommendations during sourcing events
- Risk monitoring using predictive models and third-party signals
- Contract analysis to identify obligations and deviations
- Supplier intake and onboarding automation to reduce manual data entry and accelerate supplier setup
- Conversational copilots to answer questions like “Who are our top suppliers?” or “How much do we spend with this vendor?”
On the surface, these applications appear effective. But they all depend on one assumption that rarely holds true: that supplier data is accurate, complete, and connected.
In 2024, nearly half of procurement teams (49%) piloted generative AI, and many have already seen measurable benefits. For example, productivity and effectiveness improvements have gone up by 25% in some cases.2
Despite this, meaningful outcomes still lag behind ambition. 80% of CPOs consider investing in AI a priority, and 66% consider it a high priority for the year ahead, yet only 36% of organizations report meaningful generative AI implementations today.3 These gaps reflect a broader trend: adoption without the data foundation necessary to make AI outputs reliable and actionable.
The gap isn’t caused by a lack of tools. It’s caused by data.
When supplier records are disconnected, AI cannot reliably determine which suppliers are the same, how spend rolls up across entities, or where risk truly sits. The result is AI that appears confident, but delivers incomplete or misleading outputs.
Why Supplier Data Is the Biggest Barrier to AI in Procurement
Procurement systems today are modern, but the supplier data inside them rarely is. Across industries, organizations still grapple with:
- Multiple versions of the same supplier
- Missing or outdated tax IDs, addresses, or descriptions
- Conflicting payment terms across business units
- No consistent corporate hierarchy linking local entities to global parents
- Redundant records created through unmanaged intake processes
These issues are widespread. In this report, 53% of procurement leaders rated their supplier data quality as poor, while 0% rate it as excellent. And heading into 2025, 17.6% identified unreliable supplier data as a top procurement challenge.
This creates what’s often called the hidden data factory, which is the behind-the-scenes manual work required just to make data usable. Analysts, AP staff, and category managers spend countless hours reconciling supplier lists, correcting attributes, and stitching together records before any real procurement work can begin.
And when AI enters the picture, this hidden factory doesn’t disappear. It accelerates.
At the heart of this problem is supplier identity. Most procurement organizations don’t struggle because they lack supplier data, but because they cannot reliably tell when multiple records refer to the same supplier. Variations in naming, legal entities, subsidiaries, and regional records make it difficult to establish a single, consistent view of who a supplier actually is across systems.
This challenge is known as entity resolution. In practical terms, it is the process of identifying, validating, and linking all records that represent the same supplier so spend, risk, contracts, and performance can be accurately rolled up and understood.
Why Entity Resolution Is Critical for AI in Procurement
Organizations often underestimate how deeply supplier data limits AI potential. If systems treat five variations of the same supplier as separate entities, AI will too. That directly affects:
- Spend analytics
Without resolved supplier identities, spend is split across duplicate records and subsidiaries. AI cannot accurately roll up volumes, which leads to distorted category views and missed savings opportunities. - Risk modeling
Supplier risk often exists at the parent or network level, not the local entity. When identities are unresolved, AI evaluates risk in isolation, creating blind spots and false confidence. - Supplier consolidation
AI-driven consolidation recommendations depend on knowing which suppliers are truly distinct. Duplicate or misaligned records lead to flawed rationalization strategies and unrealized leverage. - Diversity reporting
Inconsistent supplier identities cause diversity attributes and certifications to be applied unevenly. AI-generated reports then overstate or understate diversity spend, reducing trust in the data. - Category strategies
Strong category strategies require a complete view of supplier relationships across the organization. Without entity resolution, AI insights remain tactical and fail to support cross-category or enterprise-level decisions. - Contract intelligence
AI can only analyze contracts effectively when it understands which agreements belong to the same supplier. Unresolved supplier identities lead to duplicated negotiations, inconsistent terms, and limited visibility.
Without clear supplier identity, AI cannot interpret relationships, evaluate exposure, or surface meaningful recommendations.
It’s no coincidence that procurement still lags other functions in AI maturity. Research shows that procurement represents just about 6% of enterprise AI use cases, far lower than functions like sales and product management, underscoring the opportunity and the risk of moving too fast without data readiness.4
From Cleanup to Continuous Data Health
Many organizations attempt to fix supplier data reactively through periodic cleansing projects. But cleanup is a temporary reset, not a strategy.
Supplier data degrades for structural reasons. Procurement leaders report that:
- 41.2% struggle to integrate data across multiple systems
- 29.4% lack the technology to automate and maintain data quality
- 23.5% admit data quality is not prioritized internally
Without systemic change, even “cleaned” data erodes quickly.
A sustainable supplier data foundation requires:
- Clear Supplier Identity and Hierarchy
Anchoring suppliers to validated legal entities prevents duplication and supports accurate corporate hierarchies. - Continuous Refresh
Supplier information evolves constantly; data must evolve with it. - Governance Owned by the Business
Data quality isn’t just IT’s job. It’s important that procurement, finance, risk, and AP own it. - Prevention at the Point of Creation
Intake, onboarding, and M&A workflows must enforce standards upfront. - A Shared Source of Truth Across Functions
Supplier data must flow consistently across ERP, S2P, CLM, risk, and analytics systems.
The Hidden Cost of Poor Data in AI-Driven Procurement
Poor supplier data is an operational and financial issue that drives hidden costs such as:
Manual reconciliation
Teams spend significant time matching supplier records across systems just to answer basic questions. This is the hidden data factory in action, delaying analysis and limiting AI’s impact.
Repeated onboarding
When supplier identities are unresolved, records become duplicated, disconnected, and inconsistent across systems. AI cannot reliably determine which suppliers are the same, how spend rolls up across entities, or where risk truly sits.
Incorrect payment terms
Inconsistent or outdated supplier records lead to mismatched payment terms across systems. The result is overpayments, missed discounts, and disputes that erode trust between procurement, finance, and suppliers.
Missed consolidation opportunities
Without a complete view of supplier relationships, organizations fail to identify opportunities to consolidate spend. This limits leverage in negotiations and reduces the impact of sourcing and category strategies.
Unidentified risk exposure
When supplier data is incomplete or disconnected, risk signals are assessed in isolation. This prevents organizations from seeing enterprise-level exposure and increases the likelihood of surprise disruptions.
As Stefanie Fink, Head of Global Data & Digital Procurement at Kraft Heinz, noted, “There is a cost to storing bad data… It’s taking up space in your systems. It’s creating silos because everyone’s trying to do their own thing with it.”
Organizations often don’t see these costs clearly until AI initiatives stall.
The Future of AI in Procurement Depends on Data Quality
Forward-thinking procurement leaders are no longer deciding whether to use AI, they’re deciding how far they can responsibly take it.
The same AI use cases procurement teams rely on today, including spend analysis, supplier discovery, risk monitoring, contract intelligence, and conversational copilots, do not disappear in more mature organizations. They evolve.
When supplier data is accurate, connected, and continuously governed, AI moves beyond answering isolated questions and begins delivering forward-looking, network-level intelligence. In these environments, it helps them make better decisions.
As a result, leading organizations are extending today’s AI capabilities to:
- Anticipate supply risk before disruptions occur
- Identify consolidation and innovation opportunities across supplier networks
- Enable faster, more informed category strategies grounded in a complete view of supplier relationships and performance
- Support autonomous workflows once supplier identities and hierarchies can be trusted
This is the same AI applied to the same workflows, but operating with a much higher level of confidence.
Procurement teams piloting AI already report improvements in productivity, effectiveness, and user experience of up to 25%, signaling the promise of what’s possible when the foundation is right.5
But these outcomes only materialize when AI is built on accurate, structured, continuously governed supplier data. The organizations that thrive in the next era of procurement will be the ones that fix the data first, then scale the intelligence.
Sources
1State of AI in Procurement in 2025, Art of Procurement, May 2025.
2Embracing the Future: How Generative AI Is Revolutionizing Procurement in 2025, The Hackett Group, March 2025.
3State of AI in Procurement in 2025, Art of Procurement, May 2025.
4State of AI in Procurement in 2025, Art of Procurement, May 2025.
5Embracing the Future: How Generative AI Is Revolutionizing Procurement in 2025, The Hackett Group, March 2025.
Alex Denomme is a Solution Engineer at TealBook.
