For large, diversified companies, the debate over centralization versus decentralization shapes decision speed, spending, and whether technology supports or slows growth.
Paul Krebs has experienced both approaches. As CIO and chief transformation officer at Koch Industries, and previously a technology leader at The Coca-Cola Company, he has worked in settings where business units seek independence while corporate leaders push for efficiency. His perspective: the solution isn’t favoring one model, but creating a system that adapts when needed.
When decentralization becomes sprawl
Decentralization often begins as the default choice. Local teams understand their markets, customers, and regulations better than a central office. Granting them control can accelerate decisions and improve accountability. Over time, however, that autonomy may lead to duplicated systems, inconsistent data, and rising IT costs—a burden that grows with the business.
Krebs remembers Koch Industries’ digital transformation effort around 2016. Ambitions were high, but execution capabilities varied across business units. The company invested in shared services and centers of expertise—centralized teams focused on business transformation, enterprise applications, and data analytics. The aim wasn’t control, but speed.
Related: KeyBanc questions Salesforce Agentforce maturity amid weak traction
“There was always a belief that the centralization push should be re-looked at on a regular basis, not thought of as a forever decision,” he says. The strategy succeeded because it remained flexible. Some functions, like infrastructure and cybersecurity, stayed centralized. Others, such as data insights, returned to business units once local teams developed sufficient expertise.
The middle ground: standardization without rigidity
The distinction lies in knowing what must stay uniform and what can vary.
A better approach identifies the core need, builds a scalable solution, and lets teams adapt within set boundaries. This delivers both consistency and agility.
The balance becomes critical in areas like ERP consolidation. Companies built through acquisitions often end up with dozens of ERP instances. Some leaders push for full consolidation. Others prefer integration layers. If processes are truly separate, separate systems may be justified. If not, the complexity isn’t worth it.
Emerging technologies like AR/VR rarely make sense for every unit to develop its own capabilities.
Related: New Face of IT Leadership Emerges
Making the trade-offs visible
In decentralized organizations, IT costs frequently appear high compared to benchmarks. The issue isn’t always inefficiency—it’s that the trade-offs aren’t clear. Centralized services like infrastructure and security performed well on cost metrics, while decentralized areas such as business intelligence and commercial applications often had redundancy.
The answer isn’t to criticize business units for seeking flexibility. It’s to make the economics transparent. “Leaders must recognize when flexibility may cost the company, and be clear on whether the value justifies it,” he said. This changes the conversation from IT spending to business service economics. What matters is a capability-based view showing where scale works and where fragmentation increases costs.
For new CIOs entering a decentralized setting, Krebs recommends starting with curiosity. “I would begin by understanding why decisions were made,” he said. From there, the focus should align business architecture with technology, data, and organizational capabilities. A low-maturity capability might benefit from centralization to build talent and avoid reinventing solutions. As maturity improves, decentralization can provide needed flexibility. Later, the organization might centralize again to leverage scale.
The error is treating the operating model as permanent. “Once you choose centralized or decentralized, you don’t have to stay with that model,” Krebs said. The real challenge is knowing when to adjust.
