As he works to build a Department of Community Safety, Mayor Mamdani should pioneer a new way to measure not just crime, but broader public safety.
New York City is a global exporter of culture and capital (“Seinfeld”! Wall Street!), but in the world of government data nerds, one of its most enduring exports is a management tool that has been adopted across the globe: CompStat. Developed by the NYPD in the 1990s, CompStat was a deceptively simple innovation. It tracked measures of crime at the precinct level, created forums for holding commanders accountable and reinforced a basic management insight: You cannot manage what you do not measure.
The model spread to police departments all over the country and birthed “Stat” programs across federal, state and local governments for everything from pothole repair to child welfare caseloads. At their best, performance stat programs enabled a relentless focus on outcomes that energized accountability, rapid innovation and a push for results. But three decades later, as New York stands up a new Department of Community Safety, we need a new innovation. The rigor of CompStat needs to be harnessed to a new measure that can tell us more about whether city residents feel safe.
When my team interviewed mayors across the country recently, we asked each one: “What is your top priority?” The answer was almost always public safety. Next we asked, “How do you measure it?” Imagine my surprise when mayors said to us, “We don’t have a way to measure whether people feel safe. And we need one.”
CompStat measures crime reports and police activity. But it does not measure public safety directly. Mayors have dashboards filled with proxies: shootings, burglary reports, arrest rates. These are measures of system activity and system failure. They tell us how busy the police are and how often residents report that specific laws are broken. But none of them answer the fundamental question that mayors — and residents — actually care about: Do people feel safe in their city?
Without a reliable outcome metric, we pour resources into interventions without a feedback loop to tell us if they are working. And policymakers continue to make promises to historically disinvested communities without an objective measure to use to see if their promises have actually been achieved. To move to governing for outcomes, New York needs to invent a new metric — one that measures safety not merely by the absence of crime but by the presence of vibrant community life.
Safety shows up in how people act
One way of finding out whether people feel safe is to ask them. But to have a performance measure you can really manage toward, it needs to be as inclusive, real-time and easy to gather as possible. While micropolling is gaining ground in some cities, surveys and polls are usually expensive and slow to run, plus have such low response rates that they skew whose voices get heard. Is there a way to represent everyone’s experience much more equally without having to continuously ask them to respond to intrusive questions? This feels especially important to me since we’ve heard from many community members that they are frustrated and fatigued with research and “community engagement” that doesn’t lead to tangible change.
What are some other indicators that might give us real insight on how communities are experiencing safety actually looks like?
Because they’re constantly in touch with residents, mayors possess an instinct for what safety looks like outside of their dashboards. We asked them, “If you were dropped into a strange city with no access to their crime data, how would you know if it was safe?” They answered immediately that they’d look first to the use of public space. They’d want to know: Are women jogging alone after dark? Are children playing in parks? Can they ride their bikes outside in their neighborhood? Are people strolling easily, or do they walk looking over their shoulders?
Crucially, these mayors’ instincts are shared by the residents we talk with as well. In St. Paul, Minnesota, we interviewed young people from neighborhoods most impacted by violence. We asked them what they would do if their neighborhood felt safe to them. They said safety meant being able to hang out in a park, play outside or walk without scanning for threats. One 17-year-old participant told us, “I would love to just be able to go outside without worrying that something will happen to me, when someone else will just do something violent, and yeah, having walks without worrying.”
If families in a neighborhood don’t let their kids use the local park, that is a data point. If foot traffic on a commercial corridor evaporates at 8 p.m., that is a signal. These behaviors are the “revealed preferences” of public safety.
Measures of revealed preference are how we assess many other areas of urban policy. We measure the success of a transit system by ridership, not by asking people if they like the bus. We measure the vitality of a housing market by demand. We should measure public safety by the degree to which residents trust their environment enough to inhabit it.
There may be drawbacks to using perceptions of safety as a metric. What if feelings of safety are based on inaccurate perceptions of crime, for example? Survey data for the past few decades show an ongoing gap between the perceptions of crime and actual measured crime rates (though people tend to believe that crime increases are worse nationally than locally). Many city leaders right now are celebrating violent crime rates dropping to historic lows in their cities but may not see those drops leading to historic improvements in perceptions of safety (though overall perceptions may have improved somewhat in 2025.) Mayors we talked with were frustrated with resident perceptions of safety that they felt were based not on personal observations but on alarmist media, true crime shows or social media and neighborhood apps that drive engagement by amplifying a sense of vulnerability to crime.
However, the divergence between measurements of crime data and resident perceptions underlines the importance of finding a way to better understand how residents are experiencing the city. Reducing crime does not necessarily increase resident sense of safety. One reason is that many factors outside of crime may influence perceptions of safety. One mayor told us that his residents told him that one of the major factors driving their sense of public safety was the number of discarded needles in public areas. Another mentioned community members asking for help from the city with blighted lots that made the neighborhoods feel more dangerous. Many of the factors that mayors and residents described as influencing their sense of safety (the presence of individuals experiencing severe mental illness, blight, unlit walking paths etc.) sit outside the purview of law enforcement. A metric that draws on the behavior of residents to answer the question “Do people in this community feel safe to use public space here?” would allow a mayor to capture much more than crime stats alone will show.
Designing the metric
New York has a unique opportunity to operationalize this insight. The incoming administration and the new Department of Community Safety should test a “Felt Safety Index” based on observable behavioral data. Call it SafeStat.
The ultimate goal would be to build a composite index of stable indicators that track the use of public space. This might include anonymized, aggregated cellphone location data to track foot traffic patterns in parks and commercial corridors; data on playground usage; or counts of pedestrian density at key times of day. Measures that show whether residents feel safe enough to walk alone to a bodega after dark or to let their kids play outside after school.
This approach would move us from focusing lagging indicators (crime reports) to leading indicators (community vitality). The City would need to validate these measures against self-reported senses of safety and test their robustness to other inputs (weather, holidays). But if we could create reliable measures that helped us tell whether people were acting as though they felt safe, this new form of CompStat (SafeStat?) could unlock accountability and innovation in areas that have long been overlooked. If the City deploys a team of credible messengers or installs new lighting in a neighborhood, a behavioral metric would allow us to see — in near real time — if residents vote with their feet and change their behavior in those spaces. And if an area of the city reads as unsafe on SafeStat, leaders and community members would have an accountability tool that would allow them to keep pushing until they saw meaningful change.
In a recent interview, Elizabeth Glazer described the Mayor’s Action Plan for Neighborhood Safety, an initiative of the Mayor’s Office of Criminal Justice that brought together law enforcement, a wide set of City agencies and community members to codesign approaches to improving public safety. Glazer said the central question of MAP was “What’s important to neighborhoods? What’s actually driving their sense of safety, or not? And what can you actually do about it?” Glazer said, “The idea of MAP was there are so many things that compose safety, and one crucial piece of it is the experience of people living in neighborhoods. They should be not just the subjects of patrol, but should themselves be the people who are saying, ‘Look: Here are some ideas that we need to push forward in order to be safe.’” A SafeStat metric that showed whether residents felt their community was safer could allow for the structured measurement and accountability that could allow deep community engagement programs like MAP to direct more City funds to resident-centered solutions generation and testing.
Governance and trust
Is it as simple as creating the tool and deploying it? Of course not. The power to measure behavior is also the power to surveil, which makes the governance of this data an important design question. Rather than being held by the NYPD or City Hall, this data could be held in a Community Data Trust. The City could help a community or neighborhood designate a third party nonprofit entity to hold the data gathered from the community. Community members could approve a charter that would govern the mission of the trust and how to gather, store and share the data. They could elect trustees with a fiduciary duty to oversee the use of the data. The Community Data Trust could then share the data back with the City to contribute to efforts to improve public safety within the community, but with greater transparency and trust from community members. Taking the data out of the hands of the government itself would allow residents to say to the City: “Policing actions have gone up, but our data show that seniors are still too afraid to walk to the grocery store. This intervention isn’t working — we need to try something new.”
The next export
Imagine a mayor’s dashboard that displays not just the count of tragedies, but a trend line of community confidence.
If it flatlines despite a surge in police overtime, we know to pivot. Resources would finally flow toward outcomes, rewarding the organizations and leaders that actually change the way people live. New York invented CompStat, changing the way the world polices. It is time for New York to invent what comes next: a system that holds government accountable not just for controlling crime, but for cultivating the freedom to live without fear.