AI-Ready Procurement: From Data Chaos to Real-Time Control

Artificial intelligence is no longer a future topic in procurement. Leading procurement organizations use GenAI, analytics, and process intelligence to reduce costs, manage risks, and improve strategic decisions. At the same time, practice shows: Without clean data, clear governance, and a realistic target vision, AI initiatives remain stuck in pilot status.

AI in Procurement: Innovation Meets Regulation

Technology providers are continuously expanding their solutions with AI capabilities. Modern RFx tools create text drafts, consolidate responses, and highlight deviations in offers. RAG systems (Retrieval Augmented Generation) use internal contract and policy data to deliver context-specific answers.

At the same time, regulatory requirements are increasing. The EU AI Act, the Data Act, as well as GDPR and CSRD demand transparency, traceability, and high data quality. Procurement thus becomes both an innovation driver and a compliance-relevant area.

The key question is: How do companies anchor AI so that it delivers measurable value without creating additional complexity?

Spend Analytics and Classification: The Biggest Lever Often Remains Unused

The greatest optimization potential often lies in the fundamentals. Typical challenges we see with our clients:

• Multiple ERP systems without a unified data base
• Maverick buying and local orders outside the systems
• Historically grown supplier base with duplicates and inconsistent master data
• Low bundling rate and high maverick spend

Without unified spend transparency, procurement remains only partially controllable. Standards like UNSPSC or eCl@ss create a common language and enable comparable KPIs, benchmarks, and consolidated reporting.

Three Building Blocks for AI-Ready Procurement

1. Golden Supplier – Consolidating the Supplier Base

Suppliers from ERP, CRM, and financial systems are merged, duplicates cleaned up, and assigned unique IDs. The result is a consistent basis for negotiations, reporting, and compliance.

2. AI-Powered Classification

Historical order and invoice data are mapped to UNSPSC using text mining and machine learning. Human-in-the-loop processes ensure quality and acceptance. New data can be classified in near real-time.

3. Category Scorecard

KPIs per commodity group create transparency: spend volume, number of suppliers, contract and catalog coverage, and payment terms. Based on this, bundling potential, renegotiations, and preferred supplier strategies can be prioritized with solid data.

Approach: From Scoping to Rollout

Successful companies start with a clearly defined pilot project:

1. Scoping and Data Review: Target vision, KPIs, and business case are defined. Data sources are identified and checked for quality.
2. Pilot Phase: Building the Golden Supplier, classification pipeline, and initial scorecards. Typical duration: 8-12 weeks.
3. Scaling: Gradual rollout, integration into processes and systems, addition of governance and training.

Result: Depending on the starting point, savings and control effects in the high single to low double-digit percentage range are possible.

Success Factors for AI in Procurement

• Clear Governance: Define responsibilities for data quality and classification
• Defined Taxonomy: Consistently use and version standards like UNSPSC
• Scalable Data Pipelines: Automated processes for continuous data quality
• Human-in-the-Loop: Targeted manual review for critical classifications
• Early Stakeholder Involvement: Acceptance through transparency and KPI reporting

Conclusion: Data Before Algorithms

AI only unfolds its potential when data, processes, and governance are aligned. A robust master data base, consistent classification, and meaningful scorecard form the foundation for data-driven decisions in procurement.

Next Step: Structured Spend Analysis

Companies looking to determine their maturity level and identify concrete levers benefit from a structured spend analysis. Novartum develops a realistic roadmap for spend analytics, classification, and AI deployment together with procurement and IT—with a clear business case and actionable measures.