Across Rwanda’s ministries and agencies, years of digital investment have produced large volumes of data. The digitisation effort led to numerous government services being made available online, and the further development of infrastructure to support that. Even though that has been achieved an important task remains and that is the transition between owning data and governing it in a way that enables consistent use across institutions.

This shift is reflected in the establishment of the Data Governance Unit (DGU) at the National Institute of Statistics of Rwanda (NISR). The unit serves as the operational arm of Rwanda’s public sector data sharing policy, with a mandate to move the country from fragmented data systems toward a coordinated, usable data ecosystem. Since the policy is in place and the institutional structure is emerging, the question now is how this can be implemented in practice. Cenfri interviewed NISR’S Data Governance Project Lead, Maryse Gwiza, who has been seconded from the Rwanda Information Society Authority (RISA), where she was a Government Digitalisation Analyst.

Maryse works closely with institutions on the practical rollout of data governance and ultimately, data sharing. During our conversation, she emphasised that data governance is still a relatively new concept across the Government of Rwanda and that adoption will depend on demonstrating clear, practical value. Institutions need to understand how governance helps them do their work better, whether by improving decision-making, reducing inefficiencies, or enabling more reliable data use

Where things stand today

The DGU is still in its early phase, but it has made impressive progress in set up, staffing and initiation of projects. As the project lead, Maryse, explains, the focus so far has been on building the foundations required for implementation rather than driving immediate ecosystem-wide change. This includes developing governance frameworks, defining how data should be classified, setting standards for data quality, creating operational templates, selecting cataloguing tools, and building the data sharing platform itself.

This work is essential as it creates the architecture within which data can be shared across government.

The problem the policy is trying to solve

Government institutions have historically treated data as something they own rather than something they steward on behalf of the public.

In our discussion, Maryse points to a set of recurring challenges: institutions often struggle to access each other’s data, there is uncertainty around what can be shared, many organisations do not have a clear understanding of the datasets they hold, and data quality is frequently compromised by duplication or unusable formats.

The policy reframes this landscape by positioning government data as open by default, subject to classification and access controls, and by establishing institutions as stewards rather than owners of data. This changes how institutions think about control, risk, and responsibility.

What implementation requires

Maryse highlights that governance must come before sharing data. Institutions need to classify their data, define access rights, and apply agreed standards before making datasets available. Without this, sharing introduces risk rather than value.

Each organisation is expected to establish internal data teams, with roles such as chief data officers, data stewards, and data custodians embedded within existing structures. This is what anchors data governance in day-to-day operations rather than leaving it as a centralised function.

Before data can be shared, institutions need to understand what they have. This requires systematic cataloguing and classification of datasets. In many cases, this is not yet in place, which makes data discovery a prerequisite for any meaningful progress.

Data quality is key as poorly structured, duplicated, or incomplete data undermines trust in the system. If early attempts at sharing result in unreliable outputs, institutions are likely to revert to siloed practices, regardless of the policy framework in place.

The most difficult shift, however, is behavioural. Moving from a model of ownership to one of stewardship requires institutions to rethink how they approach data. This affects internal incentives, risk perception, and accountability. It is not a technical challenge. It is an organisational one.

Adoption will therefore depend on a combination of demonstrated value and policy pressure. Institutions need to see how better data governance improves their own work, while also responding to the expectations created by the national policy framework.

The next phase

The DGU is moving from internal setup to broader implementation. The next phase will involve engaging institutions directly, supporting them to establish data teams, catalogue their data, and apply governance processes in practice.

What success will look like

Looking ahead, Maryse describes success in practical terms. Over the next three to five years, institutions should have functioning data teams, a clear understanding of their data assets, and the ability to publish quality-assured datasets through the data sharing platform. This platform was built as a central system to enable institutions to publish, discover, and request datasets across government. It acts as a controlled environment where data can be catalogued, classified, and accessed based on defined rules, rather than being exchanged informally between institutions. With the platform data silos should be reduced, and shared data should be used more consistently in decision-making across government.

She says, “In practice, success will not be defined by the existence of frameworks or platforms. It will be visible in how institutions operate, specifically whether they routinely use each other’s data as part of normal processes… The DGU can set standards, provide tools, and coordinate implementation. It cannot substitute for ownership within institutions.”