From data to decisions: Improving data use in governments

Developing countries generate data by the terabyte – from taxes and trade to health, education, and infrastructure – but most of it lies unused. The real constraint is not data scarcity, but a lack of systems, incentives, and capacity to use it well. 

Administrative data – the kind generated during the delivery of public services – is particularly underleveraged. Despite its richness and regularity, it is rarely used to inform decisions. Ministries often monitor outputs but struggle to use data to shape policy, allocate resources or improve delivery. Changing that is possible. The returns would be enormous.

How can better data systems support policymaking?

Examples from many countries suggest what is possible when data is treated as a strategic asset rather than a bureaucratic by-product.

In Zambia, cancer case records were used to decide where to site new treatment centres, targeting regions with a higher disease incidence. In Kenya, digitised traffic data helped identify accident hotspots. In Rwanda, VAT data from electronic billing machines improved compliance and reduced tax leakage. In India, IGC-funded researchers are using AI models to parse tens of thousands of environmental court rulings to improve regulatory accountability.

None of these interventions required more data, just better tools and systems to use what already existed.

The real barriers: silos, skills and short-termism

Why, then, are such examples the exception?

First, most data is hard to find, access or interpret. It is stored on personal drives, locked in silos, or formatted inconsistently. Common identifiers are missing. Sharing protocols are vague or non-existent. The result is fragmentation – everyone holds a piece, but no one has the whole picture.

Second, governments’ technical and analytical capacity remains limited. Ministries rely on a handful of overburdened analysts, while officials are seldom trained – or expected – to use data in daily decision-making. The few who do are often discouraged by long lags, poor documentation, and fragile infrastructure.

Third, institutional incentives are misaligned. Political cycles discourage long-term investment in analytical capacity. Risk aversion and reputational concerns keep data locked away. Donors fund data collection or external consultants, but not the more challenging task of building internal capacity to use it where impact is harder to attribute. 

The result is a paradox: governments are awash with data, yet policy remains insight-poor.

A demand-first strategy for public sector data use

Unlocking the value of data in government is about more than better tools – it requires a fundamental shift from a ‘supply-side’ focus on generating data to a ‘demand-side’ focus on how data can be used to drive decision-making. That’s why the IGC is pursuing a demand-driven strategy with three interconnected elements.

Start from the top: Building internal capacity

Ministers and senior officials often want dashboards or decision tools that provide a clear, timely picture of what's happening. A strategy that builds the internal capacity to deliver these, rather than relying on short-lived external consultancies, can drive change across government. When decision-makers start using dashboards in one area, they are likely to demand similar tools elsewhere, expanding the appetite for clean, digitised, real-time data. As this demand grows, other officials not only become emboldened but are increasingly expected to follow.

To institutionalise this data use, civil servants will need to become intelligent consumers of evidence: able to request data and commission analyses, understand their limitations, and apply them to the decisions they face. All of this means investing in both skills and culture. In many ministries, data is still viewed as something external, produced by consultants, not as a core part of how the organisation thinks or operates. Shifting that view takes time and trust. It also takes role models: when senior leaders ask for data, others follow.

The supply side must evolve in parallel. Analysts concurrently need technical skills and the ability to engage with policy questions and work in cross-functional teams. Yet, too often, capacity-building efforts focus too narrowly on investing in supply-side factors without addressing whether there is demand for their work. The result is either skilled analysts with no clear mandate, or tools developed in isolation from policy needs.

A more effective model would ensure that capacity-building priorities are shaped by policymakers' needs, not the other way around. It would embed data work into the rhythms and pressures of public decision-making, helping move data from the sidelines into the centre of government operations.

Using the data that already exists

The priority is not to collect new data, but to unlock the value of what already exists. Governments are already sitting on rich administrative datasets, covering everything from tax compliance and labour inspections to school enrolments and procurement. The challenge lies in making this data accessible to decision-makers. 

This may require a set of technical interventions, including digitising data, standardising formats, introducing common identifiers, and building lightweight systems (which prioritise functionality over complexity) that enable real-time or near-real-time insights where these are needed. It also requires developing the capacity to produce outputs that serve immediate decision needs – whether to reallocate inspectors, adjust programme budgets, or respond to external shocks. Speed matters. So does relevance. The good news is that AI is making this easier, faster, and cheaper.

For example, in Zambia, we’re working with the Ministry of Finance to replace static reports with interactive dashboards that pull information from live data sources, enabling quicker and more informed macroeconomic monitoring. 

In Paraguay, an IGC-supported project is helping customs authorities build a machine-learning model to improve risk targeting by using existing trade data. Rather than designing new protocols from scratch, the model builds on the data they already collect every day.

Embedding support within government systems and priorities

Sustainable data use by governments cannot be outsourced. It must be built within the institutions that already deliver public services. That means aligning support not just with technical systems but also with the way government actually functions – its incentives, constraints, and priorities. 

Too often, donor-funded initiatives are designed off-site and delivered off-cycle, missing the windows when priorities are set and decisions are made. Real traction comes when support is anchored in government priorities and timelines, not external log frames. Reform processes are rarely linear. They need support models that adapt – sometimes accelerating, sometimes holding steady – as political realities shift.

This requires humility and adaptability from external actors. The job is not to dictate reform agendas, but to focus on the problems that ministries themselves are trying to solve – whether it's plugging revenue leaks, managing health system bottlenecks, or improving procurement efficiency. Quick wins help build confidence, but long-term partnerships build trust; and trust is the currency that sustains change when politics shift, or priorities evolve.

One way to institutionalise this is through embedded teams of analysts and researchers who sit within ministries, respond to real-time needs, and co-produce insights with civil servants. The Zambia Evidence Lab is one such example, offering technical expertise with political and operational alignment. Their support adapts to shifting realities because it’s grounded in them. Ultimately, embedding support means treating governments not as delivery vehicles, but as partners. It means accepting that governments must drive progress and that the legitimacy of reform rests on national priorities, not external blueprints.

On 17 June, the IGC will host a closed-door roundtable at the LSE focused on bridging the gap between data and decisions. This will be followed by a public panel co-hosted with the Data Science Institute as part of LSE Festival.

Join us at the event