A different way for cities to build data capacity

Cities are under growing pressure to make better use of data and artificial intelligence to deliver results for residents. And the dominant advice has been consistent: First, lay the groundwork by standardizing your data, setting up governance processes, and investing in central data stores. Only then, the thinking goes, can local leaders begin to roll out more ambitious, data-driven solutions on challenges from housing to public safety to infrastructure.

But the imperative to show value and improve lives in tangible ways is too urgent to wait for all of that groundwork to take hold, which is why the most effective data teams are taking a different path. They start with urgent problems, demonstrate results quickly, and develop lasting capacity along the way. In this model, capacity is not a prerequisite for action. It is a by-product of it.

Ultimately, this is how local leaders can show residents what’s possible and build their support for using data and AI in more ambitious ways in the future.

As executive director of the Bloomberg Center for Government Excellence at Johns Hopkins University (GovEx), I work every day to help cities navigate this tension of when to invest and when to act. As I do so, I lean on my own experience in city hall.

In 2010, five years after Hurricane Katrina flooded 80 percent of New Orleans’ housing stock, the city was still grappling with tens of thousands of blighted properties. These homes depressed property values, discouraged people from returning, and attracted crime. Residents who had fought insurers and navigated rebuilding programs were frustrated to see their neighborhoods still struggling despite their efforts.

When Mayor Mitch Landrieu took office that year, he set an ambitious goal: reduce blight by 10,000 addresses, about a quarter of the total, by the end of his first term. I was asked to lead a data team to support that effort.

Conventional wisdom would have suggested building strong data governance, management systems, and infrastructure before attempting to deliver results. But we didn’t have the luxury of time. The problem was urgent, and waiting years to get the foundations right would have meant years more blight, declining trust, and missed opportunity.

So we reversed the sequence. We focused on solving the problem first.

We created BlightStat, a performance management program that used data and analytics to guide city action, such as “nudging” homeowners to improve their properties after 311 complaints were filed against them. That approach yielded demonstrable results, and with that momentum, we ramped up the sophistication and impact of our data interventions. We built a machine learning-based recommendation tool to help code enforcement officials determine the most effective path for each property. We used A/B testing to understand which interventions best compelled property owners to act. And we developed civic applications that made the remediation process more transparent to residents, helping rebuild trust in city government.

The results were dramatic, and the city reached its goal well ahead of schedule.

But the more important outcome was how the work reshaped the city’s capabilities. In the process of addressing blight, we built stronger data governance, improved our data management practices, and deepened our analytical expertise.

For example, through the work on BlightStat, we created a platform called NOLAlytics, which became the foundation for projects in other policy areas, from improved smoke-alarm distribution to optimal ambulance deployment to A/B testing of texts promoting uptake of primary care. Each new application built on capabilities that had been developed because our previous work on blight had demanded them.

This is what it looks like when capacity is a by-product of solving real problems, not a prerequisite.

And yet, to this day, public-sector data programs are still designed to invest heavily in governance frameworks and technical infrastructure before delivering tangible results. There are understandable reasons why. City leaders are rightly concerned about data privacy, security, and compliance. Many operate in environments with fragmented systems, limited technical capacity, and high scrutiny around the use of data and AI. In that context, investing in governance and infrastructure upfront can feel like the safest and most responsible path.

But in practice, this approach often delays impact and makes it harder to build the momentum and trust needed to sustain those investments over time.

A problem-first strategy works differently. It generates early wins that build political and public trust. It reveals, in concrete terms, which data governance and management practices are actually needed. And it ensures that capacity is built in service of real outcomes, not abstract ideals.

This is the model Bloomberg Philanthropies has advanced through its work with cities on data and AI in the City Data Alliance: Start with a high-priority problem, build a solution that delivers results, and use that work to establish the systems and practices needed to sustain and scale.

In Santiago, Chile, for example, the municipal government struggled to keep up with waste collection. By developing a mobile application that adjusts truck routes in real time based on conditions on the ground, the city improved the accuracy of arrival-time predictions offered to residents to 90 percent. This not only built trust in municipal services, but also helped the city strengthen its data integration and operational systems.

And in Chattanooga, Tenn., leaders built a tool that connects qualified tenants with affordable housing options and providers, helping place more than 3,000 residents in permanent housing and reduce homelessness by 40 percent. Building the tool, in turn, helped the city develop new data-sharing practices and cross-agency coordination.

Both of these efforts began with a clearly defined problem that mattered to residents. Both delivered measurable results. And in both cases, the process of solving the problem clarified the governance structures, data practices, and operational systems needed to sustain and scale the work.

Cities do not need to wait to become “data-ready” before acting. In fact, the opposite is often true. By acting on urgent problems, they become data-ready.

Oliver Wise is executive director of the Bloomberg Center for Government Excellence at Johns Hopkins University (GovEx).