In college, a class opened my eyes to how policy decisions ripple through people’s lives, and it set me on a path to study these systems. My professor described how changes to the nation’s welfare program—the transition from Aid to Families with Dependent Children (AFDC) to Temporary Assistance for Needy Families (TANF)—reshaped America’s social safety net.

Today, as a researcher at NORC, I work with state agencies that administer programs such as TANF, a federal cash assistance program for families with low incomes. These agencies collect large amounts of data about the families they serve, but many struggle to turn their data into actionable insights that inform decision-making and operations.

For our work leading the needs assessment for the federally funded TANF Data Innovation project, my colleagues and I developed the first empirical framework to classify how well state TANF agencies use their data. We designed the framework as a descriptive tool to measure agencies’ variations in data practices, not as an evaluation of agency performance.

We surveyed 43 state agencies and reviewed hundreds of public documents to understand what separates high performers from the rest. Our inquiry focused solely on how agencies use data to support learning, program improvement, and evidence‑building. We shared our initial findings in 2022, and now our peer-reviewed article has been published in Poverty & Public Policy.

Using our framework, we classified agencies into three categories: basic data users (51 percent of agencies), advanced data users (21 percent), and exemplary data users (28 percent). Nearly one-third of agencies qualified as exemplary, which was encouraging; it showed that strong data use is achievable within the constraints of state government.

We defined “exemplary data use” as more than just collecting statistics or producing reports. It means agencies are analyzing data to answer questions about program effectiveness, testing innovations, and sharing findings publicly to contribute to evidence-based policymaking nationwide.

Here’s the finding that surprised us: What distinguished exemplary data users wasn’t their technology or even their staff’s technical skills—it was their culture of communication, collaboration, and transparency around data.

We found exemplary data users working with brand-new systems and exemplary data users working with systems that were over 20 years old. Surprisingly, agencies with recently upgraded systems were less likely to show strong data use in the years immediately after implementation.

This finding challenges the conventional wisdom that investing in new technology alone is the path to better data practices. Our research suggests that significant staff effort goes into implementing new systems, which can temporarily reduce attention to analytics and collaboration, the activities that drive effective data use.

The agencies with exemplary data use shared several characteristics:

• They communicate regularly about data. Exemplary agencies maintain frequent communication among staff who enter data, analysts who use it, and leadership who make decisions based on it. They also stay in contact with other state agencies such as child welfare, workforce, and Medicaid.
• They integrate diverse data sources. Eighty-three percent of exemplary users had access to more than five integrated data sources, compared to 48 percent of basic users. Multiple data sources can offer a complete picture of how families interact with programs and measure outcomes after program exit.
• They build strong external partnerships. Exemplary agencies were more likely to have data-sharing agreements with universities or research organizations like NORC, with research partnerships built on clear expectations and mutual benefits.
• They have leadership support. Exemplary data users reported greater financial resources for analytics, and their leadership regularly requested data analyses. When leaders prioritize data, staff have permission and space to do analytical work. 

TANF agencies need strong analytics to understand whether their programs are working and to test improvements before rolling them out widely. Without data analysis, agencies can’t detect unintended consequences of policy changes or learn from innovations in other states. And they can’t share evidence of what works.

When agencies use data well, they can answer critical questions: Are families who complete job training programs more likely to find stable employment? Do certain case management approaches lead to better outcomes?

For TANF directors or other leaders looking to improve their agency’s data use, our findings point to several promising strategies:

• Start with communication. Making data use a priority in organizational conversations may be a foundational, no-cost step toward improvement.
• Staff need scheduled time for analysis. Unless data analysis is part of someone’s job description and workload, analytical capacity can be limited.
• Cross-departmental communication may support stronger data use. Exemplary data users were more likely to communicate regularly with other departments.
• Thoughtful external research partnerships can help. Clear communication expectations and mutual benefit appear key to success.
• Our findings suggest agencies should carefully plan for system upgrades. Agencies may benefit from first identifying priority analytical questions. Understanding current data use helps ensure new systems support evidence-building rather than disrupting analytical capacity.

We conducted our research before artificial intelligence (AI) tools became widely available. Agencies will still need staff with time and training to apply these tools appropriately while navigating state and federal privacy regulations.

At NORC, we understand what state agencies are up against. Many of us have worked in state government or with state agencies for decades. We know the constraints agencies face, the compliance demands on their time, and the capacity issues they may face. We’re ready to partner with agencies to help fill those gaps and answer their information needs.

We know that state agencies’ top priority is ensuring they can effectively serve the families who depend on them. By using data well, they can design better programs, test innovations confidently, and contribute to a growing evidence base that helps all states learn what works.

• NORC researchers measured data use in state TANF agencies, classifying agencies as basic (51 percent), advanced (21 percent), or exemplary (28 percent) data users.
• The age of an agency’s data system shows little correlation with data use quality; recent system upgrades may temporarily reduce analytical capacity.
• Strong data use is driven by culture, not technology: frequent communication about data, integration of diverse data sources, productive external partnerships, and leadership support distinguish high performers.
• Exemplary data use enables agencies to test program improvements, identify what works, and make evidence-based policy decisions.