Data Strategy for the Canadian Public Sector: What Actually Needs to Change
Canada ranked sixth globally on the United Nations E-Government Development Index in 2003. By 2022 the ranking had fallen to thirty-second. Only 23 percent of Government of Canada services are available online end-to-end. The Dais at Toronto Metropolitan University's 2024 assessment of Canadian government digital maturity found fragmented bureaucracy, legacy systems, deficient digital skills, and equity gaps as the four structural barriers consistently limiting progress. The AccelerateGOV 2025 survey of 228 Canadian public servants found the same barriers in the same order.
The 2023-2026 Data Strategy for the Federal Public Service, now in its third year, is a serious document with a clear framework across four mission areas: data by design, data governance and management, data sharing and interoperability, and data-driven services. Year Two progress has been real in specific areas. Transport Canada launched an AI-powered Regulatory Platform in June 2025 that transforms federal acts and regulations into structured, searchable data. The GC Data Community is building a common data maturity assessment tool. Indigenous data sovereignty work is progressing through the Transformational Approach to Indigenous Data initiative. Nearly 100 active external data requests are being tracked with formal information sharing agreements for 40 of them.
The honest assessment is that these are genuine achievements within a system that remains structurally constrained. The progress is real and the trajectory is positive. The pace is not matching the urgency of the Carney government's transformation agenda, the scale of AI investment that federal departments are being asked to absorb, or the expectations of Canadians who are experiencing the gap between private sector digital services and government digital services every time they interact with a federal program.
What follows is a frank diagnosis of where public sector data strategies succeed and fail, what the specific design choices are that produce durable progress versus compliance documentation, and what the Carney government's political moment means for organizations that are in a position to move faster than the traditional pace of government transformation allows.
Why Government Data Strategies Underperform
Public sector data strategies fail in predictable ways that are worth naming precisely, because the organizations that avoid these failure modes are the ones producing the genuine progress that the GC Data Strategy's Year Two report documents, and the ones that fall into them are the organizations whose data strategies remain strategy documents rather than operational realities.
Treating Data Governance as Compliance Rather Than Operations
The most common failure mode in government data strategy is designing data governance as a documentation and compliance exercise rather than as an operational capability. Governance that produces policies, frameworks, and role definitions without changing how data is actually managed, shared, and used in day-to-day program operations is governance that exists on paper and nowhere else.
The distinction shows up clearly in the Year Two progress report. The accomplishments that represent real operational change, Transport Canada's Regulatory Platform, the active data sharing agreements with measurable external request volume, the GC Digital Talent platform that streamlines cross-government talent sharing, all involve systems and processes that people use every day to do their work differently. The accomplishments that represent governance infrastructure, policy frameworks, reference standards, guidance documents, are prerequisites for operational change but are not operational change in themselves. The departments that have moved furthest are those that connected governance work directly to operational redesign rather than treating governance as a deliverable separate from operations.
Underestimating the Departmental Sovereignty Problem
The government architecture post in this series describes the fundamental structural constraint on GC data strategy: the federal government is organized into legally independent departments with their own ministerial accountability structures. That accountability structure produces organizational incentives that are aligned with departmental data management rather than enterprise data management. A department that builds its own data platform, its own governance structure, and its own analytics capability is meeting its mandate. The fact that this replicates what twenty other departments have built is not primarily that department's accountability problem.
Data strategies that attempt to overcome this constraint through mandate and policy authority alone consistently underperform. The mandate exists: Treasury Board's Policy on Service and Digital and the associated directives create real obligations for departments. The enforcement mechanism is weak relative to the organizational incentives pulling in the opposite direction. The data strategies that produce results in this environment work with the departmental incentive structure rather than against it, giving departments a compelling answer to the question of what they gain from investing in enterprise data capabilities rather than building departmental ones.
The GC Data Community's work on a common data maturity assessment framework is a promising mechanism because it gives departments a tool for measuring and demonstrating their own progress while contributing to enterprise comparability. That combination, departmental value with enterprise benefit, is the design pattern that works in the federal context.
Legacy System Reality vs Transformation Ambition
Canada's government legacy systems are not merely aging technology. They are decades of policy logic, program rules, and exception handling that are not fully documented anywhere and that cannot be replaced without understanding exactly what they do. The GC White Paper on Service and Digital Target Enterprise Architecture describes the problem precisely: the life cycle evolution of individual systems tended to limit their scope to those individual systems, reinforced by a desire to restrict procurement, technical, and change complexity and risk.
Data strategies that require legacy systems to be modernized before data programs can produce value will wait indefinitely. Data strategies that extract value from legacy systems while building the new capabilities alongside them are the ones that produce results within political and funding cycles. The practical design principle is to build data access layers that make legacy data usable for analytics and AI programs without requiring the legacy systems themselves to change, while simultaneously building the governance and quality infrastructure that will support the eventual system replacements when they occur.
This is the pattern behind Transport Canada's Regulatory Platform. Rather than waiting for regulatory information to be restructured within legacy systems, the platform extracts the data, transforms it into structured format, and makes it searchable and analyzable. The legacy system continues to be the system of record. The platform creates a data layer that is fit for modern use without requiring the legacy infrastructure to change. That approach is replicable across many GC data programs where the value is in making existing data usable rather than in creating new data.
The Four Structural Changes That Produce Durable Progress
Data Literacy as a Prerequisite for Everything Else
The data literacy post in this series argues that literacy programs fail when they are designed to transfer knowledge rather than change behaviour. That argument applies with particular force in the public sector, where the population of people making decisions that depend on data quality and data access is enormous, the variation in current data literacy across the workforce is wide, and the organizational incentives for using data in decision-making are less direct than in the private sector.
The GC's investment in data literacy through the Canada School of Public Service and departmental learning programs is genuine. The gap, which the Year Two progress report acknowledges as an ongoing challenge, is that literacy training that is not connected to the specific data systems and decision workflows that public servants use every day does not change how those public servants use data. The departments producing measurable behaviour change are those that pair literacy training with immediate application to real decisions using real GC data, not theoretical datasets, and that build performance expectations around data-informed decision-making into the management frameworks that determine how public servants are evaluated.
Data Sharing Infrastructure as a Platform Investment
The data sharing mission in the GC Data Strategy is the area with the most direct connection to service improvement for Canadians, and the area where the structural constraints are sharpest. Citizens experience government services as a single relationship with the Government of Canada. Government manages those services as dozens of separate departmental programs, each with its own data systems, each collecting similar or identical information from the same citizens, each operating under legal authorities that restrict what it can share with adjacent departments.
The Pan-Canadian Interoperability Roadmap and the GC's enterprise data sharing work address this through standards and consent frameworks. The implementation gap is significant: the data sharing agreements and technical integration work required to actually connect departmental data for service improvement purposes is slow, expensive, and requires sustained political sponsorship to overcome the legal and organizational inertia that accumulates between any two departments trying to share data for the first time.
The organizations making visible progress are treating data sharing infrastructure as a platform investment rather than a project-by-project integration effort. Rather than building a bilateral integration between two specific departments for a specific program purpose, they are building the consent management infrastructure, the API standards, and the data access governance that allows multiple bilateral integrations to be built more quickly on a common foundation. The investment is larger upfront and produces compounding returns as each additional data sharing relationship builds on existing infrastructure rather than starting from scratch.
AI Readiness as the Near-Term Driver of Data Quality Investment
AI programs are the most politically compelling driver of data quality investment in the public sector right now, and the departments that recognize this are using AI program objectives to justify data infrastructure investments that would not have been approved as standalone data quality programs.
The logic is straightforward. A department that wants to deploy AI for service improvement, predictive analytics, or regulatory efficiency needs data that is consistent, complete, and structured enough to support model training and inference. If that data does not currently meet those requirements, the AI program creates a funded mandate for data quality remediation that the data team has been unable to secure on its own merits. The AI program is the business case for the data investment, not a separate initiative that depends on data being ready.
The Directive on Automated Decision-Making, which requires impact assessments for AI systems used in federal program decisions, creates a parallel governance pressure: departments that want to use AI in program delivery need to demonstrate the quality and lineage of the data feeding their models. That regulatory requirement is driving data documentation work that produces the lineage traceability and quality certification infrastructure that the data strategy has been trying to build through governance policy alone.
Talent Architecture That Matches the Work
The Year Two progress report identifies data talent as a continuing constraint across the GC. The GC Digital Talent Platform is a real improvement in how the government recruits and shares digital skills. The structural constraint it does not resolve is the compensation gap between what the government can offer data scientists, data engineers, and AI specialists, and what those professionals can earn in the private sector.
The public sector organizations making progress in this environment are not trying to win the compensation competition. They are making a different offer: mission-driven work at national scale, a policy environment where data insights actually influence programs that affect millions of Canadians, and job security and work environment characteristics that are genuinely differentiating relative to the private sector. That offer is compelling for a specific population of data professionals. The talent strategy that works in the public sector is one designed to attract and retain that specific population rather than to compete broadly across the data talent market where the government cannot win on compensation.
The GC's apprenticeship and co-op programs are an underutilized pipeline. Universities and colleges across Canada produce data professionals who have not yet developed the private sector compensation expectations of mid-career professionals, who often have genuine interest in public service work, and who can develop deep GC-specific domain knowledge that makes them more valuable over time in a government context than they would be as generalists in the private sector. Investing in developing this pipeline is a longer-term play than hiring experienced professionals, and it is the sustainable talent strategy for organizations that cannot compete on compensation.
What the Current Political Moment Means for Data Leaders
The Carney government's combination of significant spending reduction commitments and explicit transformation ambition creates an unusual political environment for GC data leaders. Spending reductions in a traditional fiscal consolidation would depress data investment. Spending reductions accompanied by a genuine mandate for digital transformation create a different dynamic: departments that can demonstrate cost savings and service improvement through better data use have a political opening for investment that is harder to secure in normal times.
The former CDO of Transport Canada and other public sector transformation practitioners have called for a Digital Government Executive Office with Cabinet mandate and delivery authority, precisely because the current political energy for transformation is not matched by the delivery infrastructure that can translate that energy into results within a political term. Whether or not that specific institutional recommendation gains traction, the directional point is correct: political moments that create transformation mandates are time-limited, and the data programs that are positioned to move during the current window will produce results that departments that wait will not.
For data leaders in federal departments and agencies, the practical implication is to identify the data programs that can produce demonstrable service improvement or cost reduction outcomes within 18 to 24 months, build those programs no