The CDO Role in the AI Era: Where It’s Breaking, and How to Fix It

The Chief Data Officer role was never meant to stay static, but what’s happening right now is a step change.

In financial services and consulting firms, the expectations have expanded faster than the operating model has caught up. You are still accountable for governance, quality, and regulatory alignment. At the same time, you are expected to lead AI strategy, unlock measurable business value, and enable innovation at speed.

Those things don’t naturally coexist. That’s where most CDOs are feeling the strain.

Where Things Are Breaking

1. AI ambition is outpacing data reality
Every leadership conversation includes AI. Boards want to see transformation, not incremental improvement. Clients expect intelligent products and personalized experiences.

But underneath that ambition, the fundamentals are uneven. Data is fragmented across systems, definitions are inconsistent, and ownership is often unclear. In many firms, critical data elements still move through manual processes or brittle pipelines.

The result is predictable: AI initiatives stall, not because the models are weak, but because the inputs are unreliable. And the accountability lands with the CDO, even when the root cause sits across multiple functions.

2. The risk bar is rising at the same time as speed expectations
In financial services, you are not just building models; you are building models that must be explainable, auditable, and defensible.

Regulators are increasingly focused on AI, especially around bias, transparency, and decision accountability. Model risk management frameworks were not originally designed for the complexity of modern AI, yet they are being stretched to cover it.

At the same time, the business is pushing for faster delivery. That tension is real. Move too slowly, and you lose competitive advantage. Move too quickly, and you create regulatory and reputational risk.

3. The operating model is misaligned
Most organizations still separate data, technology, and business ownership in ways that made sense five years ago, but not today.

Data teams sit centrally. Engineering may be distributed. Business units own outcomes but not the underlying capabilities. Risk and compliance operate as oversight rather than integrated partners.

The CDO is expected to connect all of this, often without direct control over funding, talent, or priorities. Influence becomes the primary tool, which is not always enough when incentives are pulling in different directions.

4. Talent gaps are structural, not just numerical
There is a lot of focus on hiring data scientists and AI specialists. That is necessary, but not sufficient.

The bigger gap is in data engineering, platform architecture, and AI governance. These are the capabilities that allow organizations to move from isolated use cases to scaled systems.

There is also a translation gap. Very few people can move fluidly between technical detail and business context. Without that bridge, even strong teams struggle to align on what matters.

5. Pilot purgatory is still the default
Many organizations have dozens, sometimes hundreds, of AI use cases. Some of them are genuinely high quality.

But very few make it into production in a way that is stable, repeatable, and embedded in core workflows.

In consulting firms, this often shows up as strong client pilots that never scale across accounts. In financial services, it shows up as models that work in controlled environments but fail to integrate into decisioning systems at scale.

6. Value is not clearly defined or tracked
There is often a disconnect between activity and impact.

Teams can point to models built, data assets created, or platforms implemented. But tying those efforts to revenue growth, cost efficiency, risk reduction, or customer outcomes is less consistent.

Without a clear value framework, it becomes harder to prioritize, harder to defend investment, and harder to scale what works.

What Actually Works

This is where I see leading CDOs starting to differentiate.

1. Treat data like a product, with real accountability
This is more than a mindset shift. It is an operating model shift.

Data products should have defined owners, clear consumers, and measurable performance expectations. Quality, availability, and usability should be managed with the same discipline as any customer-facing product.

In financial services, structuring around domains like customer, risk, transactions, and fraud creates reusable assets that can support multiple AI use cases without constant rework.

2. Redesign governance as an enabler, not a blocker
Governance needs to move closer to the point of creation.

That means embedding controls into pipelines, automating lineage and documentation, and integrating model validation into the development lifecycle.

Rather than treating governance as a stage gate, it becomes part of how systems are built. This reduces friction and increases consistency.

For consulting firms, this also becomes a differentiator; the ability to deliver AI that is not just innovative, but trusted.

3. Invest heavily in the data and AI platform layer
Platform decisions are strategic decisions.

Modern architectures, whether lakehouse, hybrid, or otherwise, need to support scalability, interoperability, and real-time access. Metadata, lineage, and observability are not secondary concerns; they are core capabilities.

This is where a lot of organizations underinvest, and it shows up later as complexity and cost.

4. Shift from projects to products and from use cases to capabilities
Instead of funding isolated use cases, leading organizations are building capabilities that can be reused.

For example, rather than building multiple independent models for customer insight, they build a shared feature store, standardized pipelines, and reusable model components.

This reduces duplication and accelerates delivery over time.

5. Build cross-functional AI pods with end-to-end ownership
Small, focused teams that include data engineering, data science, domain expertise, and risk partners can move faster and with more alignment.

These teams should own the full lifecycle; from problem definition to deployment and monitoring.

It changes accountability. It also reduces the friction that comes from handoffs between teams with different priorities.

6. Implement a disciplined value management framework
Every AI initiative should be tied to a clear business outcome.

Leading CDOs are defining value at the outset, tracking it through delivery, and revisiting it post-deployment. They are also segmenting initiatives; experimental, scaling, and enterprise.

This creates transparency and supports better decision-making about where to invest and where to stop.

7. Prioritize adoption and behavior change
Even the best models fail if they are not used.

Adoption requires more than deployment. It requires integration into workflows, clear communication of outputs, and alignment with how decisions are actually made.

In financial services, this might mean embedding models directly into underwriting or fraud workflows. In consulting, it might mean integrating AI into how teams deliver client work, not just as a side capability.

8. Elevate the role of the CDO as a business leader, not just a functional head
The most effective CDOs are deeply connected to business strategy.

They are not just responding to requests; they are shaping where the organization invests, which capabilities matter, and how success is measured.

That requires credibility across both technical and business domains, and a willingness to engage in trade-off decisions.

The Opportunity

The role of the CDO is becoming more central, not less.

In financial services, the opportunity is to turn data into a competitive advantage while maintaining trust, compliance, and resilience.

In consulting firms, the opportunity is to move from bespoke analytics to scalable, repeatable AI-enabled offerings that drive consistent client impact.

The gap right now is not in vision. Most organizations know where they want to go.

The gap is in execution; in aligning operating models, building the right foundations, and staying disciplined about value.

That is where the CDO can make the biggest difference.

Not by trying to do everything, but by focusing on the few things that actually unlock scale.