Our discussion reinforced the fact that advancing public data systems requires a coordinated approach that integrates three components: immediate technical enhancements, medium-term structural reforms, and long-term ecosystem development, specifically:

1. Immediate technical improvements include better data standardization, automated quality checks, and user-friendly interfaces that reduce barriers to access and interpretation. Tools like NORC’s Live Crime Tracker, AH Datalytics’ Real-Time Crime Index, and NYU’s Jail Data Initiative demonstrate what’s possible when technical excellence meets user needs.
2. Medium-term structural reforms mean addressing procurement rules, vendor lock-in, and funding models that currently prevent agencies from adopting better systems. This may require state and federal policy changes that treat data infrastructure as essential for public infrastructure.
3. Longer-term ecosystem development involves creating sustainable relationships between data producers and users, ensuring that agencies providing data receive value in return, and building the analytical capacity that turns raw information into actionable intelligence.

One of the most striking insights from our discussions was the mismatch between who we think needs crime data and who actually uses it. While policymakers are often seen as the primary audience, participants noted that journalists, researchers, and community organizations are far more likely to engage with data directly. Policymakers, meanwhile, tend to rely on those intermediaries to translate numbers into actionable insights.

This disconnect has profound implications for how we design data systems. Much of our infrastructure development focuses on providing more granular data rather than better interpretation and contextualization. The pattern extends to law enforcement agencies themselves, many of which would benefit from trusted data advisors who could help them synthesize and understand their own data, communicate it effectively, and use it to improve outcomes.

Our Innovation Day surfaced a critical tension in real-time crime data: the faster you release information, the greater the risk of misinterpretation. Participants emphasized that without proper framing, even accurate data can mislead the public and distort policy decisions. Law enforcement leaders shared concerns about how quickly incomplete or decontextualized information spreads on social media, often creating public pressure before agencies can provide proper context.

To support informed decision-making, it is essential to release data that are both valid and reliable, not just put out whatever data are available. Valid data must accurately reflect the intended time period, while reliable data are collected consistently using the same systems and methods. Timeliness matters, but not at the expense of accuracy and reliability.

The discussions centered on structural challenges that rarely make headlines but fundamentally limit progress. Record management system (RMS) vendors have emerged as a critical bottleneck. Many police agencies don’t actually control their own data due to vendor contracts and technical limitations. Procurement rules compound the problem, making it difficult for agencies to change systems even when better options exist.

Meaningful progress may require state-level standardization efforts, pointing to Michigan’s successful statewide approach as a model. The Michigan State Police Department established a statewide, integrated system that unifies data sources across law enforcement agencies, enabling real-time search and coordinated decision-making. However, such solutions may not work for larger, local departments with complex systems where change costs become prohibitive.

Several participants noted that agencies are being pitched sophisticated predictive models, but many lack the capacity to use them effectively. Rather than viewing artificial intelligence (AI) as primarily an analytical tool, discussions highlighted its potential to reduce administrative burden—the unglamorous but critical work that often prevents better data sharing.

Law enforcement participants were particularly interested in AI’s ability to extract structured information from narrative reports, potentially reducing the time officers spend on data entry. AI that reduces administrative burden provides immediate value while improving data quality for everyone downstream, offering a more practical approach than complex forecasting models.

The current landscape suffers from a misalignment of incentives. Agencies are hesitant to invest in improvements because the benefits are broadly distributed while costs are concentrated. Those who do invest often end up subsidizing others who don’t, creating a principal-agent dilemma. There’s also fear that innovative data use could backfire, potentially harming the agencies that provided the data.

A recurring theme was the need to provide tangible value back to the agencies providing data. Successful systems should provide analytical capacity back to contributors rather than simply collecting information. Participants discussed “low-cost solutions” like automated scripts that help agencies understand their own trends and patterns, creating a virtuous cycle where data sharing directly benefits contributors.

Throughout the day, discussions surfaced several testable hypotheses and research ideas, including the potential for real-time data to improve public safety responses, the use of AI to enhance unstructured police data utility, and the role of transparent, contextualized data in building public trust. Participants also discussed integrating crime data with socioeconomic indicators for better targeted policies.

There are already promising examples of what this ecosystem could look like. In Baltimore, integrated data systems have helped align public safety and public health efforts. In Ohio, statewide standards have improved data quality and comparability. And in public health, tools like Google Trends have been used to track flu outbreaks in near real time, offering a model for how crime data might be used to anticipate and respond to emerging issues.

The volatility of recent crime trends, rising dramatically in 2020, then declining sharply in 2023 and 2024, underscores the urgency of better data infrastructure. When crime patterns change rapidly, policies based on outdated information can waste resources and miss emerging opportunities for intervention.

Our Innovation Day discussions revealed that the technical challenges, while significant, may be easier to solve than the structural and incentive problems that currently limit progress. Building a smarter crime data ecosystem isn’t just about better algorithms or faster updates; it’s about creating sustainable relationships between the people who generate data and those who need to use it to drive research forward.