The above post recently popped up in a discussion stream, and I must admit I was a bit surprised because I had forgotten about it, even though it triggered a now-memorable debate back in the day.
Curious, I wondered, “Is the shared services model still being used in 2025?” I ask that question because, in 2008, it was a somewhat contentious issue, as illustrated by the Barry Bonds post.
Another question we must ask is how Agentic AI impacts the shared services model. For example, here are some of the purported risks associated with introducing Agentic AI into the shared services model:
Challenges and Risks
High Initial Investment: Implementing Agentic AI costs $100K-$1M+ (e.g., AI platforms, training), delaying ROI by 6-12 months vs. traditional SSCs.
Data Dependency: Effectiveness relies on clean, real-time data; poor inputs reduce accuracy by 10-20% (e.g., missed fraud flags).
Ethical and Control Risks: Autonomous decisions (e.g., supplier cuts) may conflict with policies, risking 5-10% compliance issues if unchecked, per Gartner (2024).
Workforce Disruption: Automating 70-90% of tasks could displace 20-30% of SSC staff, requiring reskilling ($50K-$200K cost), per SSON (2024).
Over-Reliance: Over-dependence on AI might erode human oversight, echoing Cummins’ “mutual delusion” if outcomes falter (5-10% risk).
Regarding point 4, beyond the dramatic SLA improvement and significant cost savings that the 1998 RAM model had on the DND, the FTE equivalent resource allocation across partnered companies saw a reduduction from 23 to 3 within 18 months.
While it is estimated that 80 to 85% of private sector and 65 to 75% public sector shared services initiatives are successful, why do ProcureTech initiatives continue to struggle with a generational failure rate of 60 to as high as 80%?
Private: Unilever’s $1B-$10B SSC with 25-35% AI savings; Public: UK GSS £600M since 2016 (Procurement Magazine 2024).
Hansen’s RAM (9/10) vs. S2P bloat ($1M-$5M vs. $10M-$20M); 30-50% failure from adoption gaps (Ardent 2023).
Mitigation Potential
Already high success; Agentic AI adds 5-10% (e.g., 85-90% private maturity, SSON 2024).
Agentic AI and I&O boost success to 70-80%, cutting non-usage to 5-15% (Zip 2024), narrowing gap with SSCs.
Beyond asking why organizations don’t use a shared services implementation model for ProcureTech initiatives, a bigger question is what are the similarities between using a shared services implementation model and my championed Agent-based solution development model. In short, is the Agent-based model a critical strategy for achieving ProcureTech implementation success?
The following is a high-level answer to the above question regarding the relationship between shared services and the Agent-based models:
The Shared Services implementation model and the Agent-based implementation model championed by Jon W. Hansen—specifically his Relational Acquisition Model (RAM)—share several common links that underpin their effectiveness in procurement and operational contexts.
Shared Services is a centralized organizational strategy for consolidating support functions (e.g., procurement, finance) with a proven success rate (80-85% private, 65-75% public), while Hansen’s RAM, an agent-based approach from the late 1990s updated in 2025, leverages intelligent agents (e.g., early AI) to optimize procurement with a lean, relational focus (9/10 rating). Despite their differing scopes—Shared Services as a broad structural model and RAM as a procurement-specific, tech-driven framework—their overlap lies in efficiency, simplicity, usability, and practical outcomes.
In my next post, I will explore these common links based on my past research and writings (Procurement Insights, 2025) dating back to 1998 and Shared Services insights (e.g., Deloitte, SSON).
Procurement Insights
What do Shared Services and the Agent-based model have in common and why it is important to procurement
Posted on April 2, 2025
0
In March 2008, I wrote the following post – and yes, in my younger years, I had a bit more of a whimsical tone than I perhaps do today: Shared Services Simplified or How Barry Bonds’ Bat Weight Explained GoC Thinking.
The above post recently popped up in a discussion stream, and I must admit I was a bit surprised because I had forgotten about it, even though it triggered a now-memorable debate back in the day.
Curious, I wondered, “Is the shared services model still being used in 2025?” I ask that question because, in 2008, it was a somewhat contentious issue, as illustrated by the Barry Bonds post.
Another question we must ask is how Agentic AI impacts the shared services model. For example, here are some of the purported risks associated with introducing Agentic AI into the shared services model:
Challenges and Risks
Regarding point 4, beyond the dramatic SLA improvement and significant cost savings that the 1998 RAM model had on the DND, the FTE equivalent resource allocation across partnered companies saw a reduduction from 23 to 3 within 18 months.
While it is estimated that 80 to 85% of private sector and 65 to 75% public sector shared services initiatives are successful, why do ProcureTech initiatives continue to struggle with a generational failure rate of 60 to as high as 80%?
Beyond asking why organizations don’t use a shared services implementation model for ProcureTech initiatives, a bigger question is what are the similarities between using a shared services implementation model and my championed Agent-based solution development model. In short, is the Agent-based model a critical strategy for achieving ProcureTech implementation success?
The following is a high-level answer to the above question regarding the relationship between shared services and the Agent-based models:
The Shared Services implementation model and the Agent-based implementation model championed by Jon W. Hansen—specifically his Relational Acquisition Model (RAM)—share several common links that underpin their effectiveness in procurement and operational contexts.
Shared Services is a centralized organizational strategy for consolidating support functions (e.g., procurement, finance) with a proven success rate (80-85% private, 65-75% public), while Hansen’s RAM, an agent-based approach from the late 1990s updated in 2025, leverages intelligent agents (e.g., early AI) to optimize procurement with a lean, relational focus (9/10 rating). Despite their differing scopes—Shared Services as a broad structural model and RAM as a procurement-specific, tech-driven framework—their overlap lies in efficiency, simplicity, usability, and practical outcomes.
In my next post, I will explore these common links based on my past research and writings (Procurement Insights, 2025) dating back to 1998 and Shared Services insights (e.g., Deloitte, SSON).
30
Share this:
Related