What the Private Sector Can Learn from Public Sector Data Leadership

In advance of CDAO Canada in Toronto, Celio Oliveira, the newly minted Chief Data Officer at the Department of Finance Canada, shares why trust, accountability, and outcomes, not technology, define data and AI success in the modern era

In commercial enterprises, the success of data and AI initiatives is often measured by their impact on the bottom line. In government, the equation is fundamentally different.

For Celio Oliveira, Chief Data Officer at the Department of Finance Canada, success is not defined by whether a product is launched, or even how sophisticated it is. It is defined by whether it improves people’s lives.

“We cannot consider a successful project only because a product is launched. If it’s not making anyone’s life easier, it’s not a success,” he says.

This reframing defines a more complete model of value, one that spans decision quality, service outcomes, trust, institutional resilience, and operational improvement.

Trust by design

Too often, trust in data and AI systems is treated as a natural result of good data practices. In reality, Oliveira argues, it must be intentionally designed.

One of the biggest barriers in the public sector is the perception, and often the reality, that citizens do not control their own data. Errors in government records can be difficult to correct, and the process is rarely transparent.

This creates a disconnect between the people and their government.
When people feel they lose ownership of their data, they become reluctant to share it, undermining the very systems designed to serve them.

For private sector organizations, particularly in regulated industries like financial services and healthcare, this dynamic is strikingly familiar. Customer trust erodes when systems feel opaque, or when individuals lack visibility and control over how their data is used.

At its core, trust is shaped by how much visibility and control people feel they have over their own data.

More technology does not mean more value 

Despite the rapid rise of AI, Oliveira cautions against equating technological advancement with progress.

Too often, organizations focus on deploying tools, whether generative AI or advanced analytics platforms, without addressing the underlying foundations required to make them effective.

“It’s not about having the best technology,” he explains. “It’s about executives having the knowledge to challenge it.”

Without sufficient data literacy and governance, even the most advanced tools can lead to poor outcomes. Vendors may overpromise, leaders may lack the confidence to question outputs, and projects risk becoming unused investments.

This is not a public sector problem alone. Across industries, organizations are grappling with the same challenge: how to ensure that technology delivers value, rather than complexity.

The answer lies not in the stack, but in the capability of the people using it.

Making decisions based on evidence 

The ambition to be “data-driven” is now commonplace. But turning that ambition into reality is still a challenge for many organizations.

Oliveira draws a distinction between data-driven and evidence-based decision-making. The latter requires more than access to data, it demands structured frameworks, robust analysis, and clear narratives.

Leaders must understand not just what the data says, but why it matters. This includes distinguishing between correlation and causation, understanding historical trends, and building forecasts that can guide action.

Without this rigor, decision-making remains reactive and incomplete.

For private sector leaders, the takeaway is significant. Evidence-based decision-making is not a technical upgrade, it is an organizational capability, one that depends on both analytical depth and communication clarity.

Governance isn’t bureaucracy, it’s the operating model

Few topics in data generate as much resistance as governance. Yet Oliveira reframes it not as a constraint, but as an enabler.

“Governance is not about being pretty, it’s about being effective.”
At its core, governance provides the structure needed to reduce ambiguity and ensure accountability. It answers critical questions: Who owns the data? Who is responsible for quality? Who approves access? Who resolves disputes?

Far from slowing organizations down, these guardrails enable better, faster decisions by creating clarity and consistency.

For private sector organizations, this represents a shift in mindset. Governance should not be viewed as overhead, but as infrastructure, essential for scaling data and AI responsibly.

Responsible AI means accountable AI

As AI adoption accelerates, the question is no longer whether organizations can deploy it, but whether they can govern it.

In the public sector, accountability remains non-negotiable. Even when AI systems are involved, responsibility cannot be delegated.

“Public institutions remain accountable for decisions, even when technology is involved,” Oliveira points out.

Oliveira outlines six guiding principles for responsible AI: purpose clarity, proportionality, human accountability, traceability, rights protection, and institutional readiness.

The underlying message applies to public and private sector organizations alike. They must be able to explain how systems work, justify their outcomes, and stand behind them when challenged.

For private sector leaders who face increasing regulatory scrutiny and reputational risk, this approach is crucial.

A converging future for data leadership

While public and private sector organizations have historically operated with different priorities, those boundaries are beginning to blur.

The public sector offers a model grounded in trust, accountability, and long-term outcomes. The private sector brings speed, innovation, and scalability.

The future of data leadership lies in combining these strengths.

__

Explore these issues further in Oliveira’s forthcoming book, Constitutional Intelligence: A Decision Architecture for Trustworthy AI Governance, which argues that trustworthy AI begins not with tools, but with the design of accountable decision systems.