What Works in American Policing Part 4—Problem-Oriented and Intelligence-Led Policing: Diagnosing the Problem Before Deploying the Response

This is the fourth in a seven-part series examining the policing strategies that shape public safety in America, including what the evidence supports, what has fallen short, and what policymakers and practitioners should prioritize going forward.

Knowing where crime concentrates is only part of the picture. Problem-oriented policing (POP) and intelligence-led policing (ILP) ask a more demanding question: Why does crime persist in particular places, and who is driving it?

• POP is problem-centered, starting with a recurring condition and working outward to identify causes and design responses that may involve multiple agencies and non-enforcement tools.
  • A POP approach to a drug market might analyze the environmental conditions sustaining it and design a multi-agency response that includes code enforcement and social services alongside targeted enforcement.
• ILP is threat-centered, beginning with the identification of prolific offenders or criminal networks and working inward to develop intelligence that guides enforcement and disruption.
  • An ILP approach to the same drug market would focus on identifying key players, mapping supply chains, and developing the intelligence needed to dismantle the operation.

In practice, the most effective agencies draw from both strategies.

Problem-Oriented Policing: The Case for Diagnosis

In 1979, University of Wisconsin law professor Herman Goldstein published a foundational critique of American policing. His argument was deceptively simple: Police had become so consumed with responding to individual incidents that they lost sight of the underlying conditions generating those incidents. A department might respond to dozens of robberies at the same convenience store without ever asking what made that location a persistent target. Goldstein called this the “means over ends syndrome” and proposed POP as an alternative. Under this framework, police would systematically identify recurring problems, analyze root causes, develop tailored responses, and measure results.

In 1987, John Eck and William Spelman piloted POP with the police department in Newport News, Virginia, through the Police Executive Research Forum. Their work produced the SARA model—Scanning, Analysis, Response, Assessment—a four-stage process for moving from observation to tailored intervention. The pilot produced measurable results, including a 39 percent decrease in downtown robberies, a 35 percent reduction in apartment burglaries, and a 53 percent decline in thefts from vehicles parked outside a manufacturing plant.

What makes POP distinctive is that it treats the problem as the fundamental unit of police work instead of focusing exclusively on the incident, location, or offender. Responses that emerge from a genuine SARA process often include code enforcement, improved street lighting, landlord accountability, or environmental redesign, shifting the police role from sole responder to lead coordinator of a multi-agency effort.

The Evidence Base

The National Institute of Justice’s CrimeSolutions clearinghouse rates problem-oriented policing as “Promising” for reducing crime and disorder. A systematic review synthesizing 34 eligible studies found that POP interventions produced a statistically significant 33.8 percent reduction in crime and disorder relative to controls, with no evidence of significant displacement and some evidence of diffusion of benefits. Across the full body of evidence, 31 of the 34 studies showed effect sizes favoring treatment, and the 2018 National Academies report on proactive policing reached the same conclusion: POP is an effective short-run strategy for reducing crime and disorder. In the Jersey City violent crime places experiment, the 28 interventions at treatment sites included requiring store owners to clean storefronts, removing trash, improving lighting, demolishing abandoned buildings, and connecting homeless individuals to treatment (alongside targeted enforcement). That mix of enforcement and non-enforcement responses is what distinguishes POP from simply increasing police presence.

The Implementation Gap

While the evidence for it is strong, POP’s implementation record is more complicated. Researchers have documented a persistent “disconnect between principles and practice” where officers frequently engaged in “shallow” problem solving—cursory analysis followed by familiar enforcement responses, with little genuine assessment of underlying conditions. Survey results found that only 33 percent of U.S. departments actively encouraged problem-solving projects, though the figure rose to 74 percent among agencies serving 100,000 to 249,000 residents. Research from England and Wales has identified structural barriers, such as competing priorities, insufficient analytical capacity, weak interagency partnerships, and a culture that rewards arrest counts over the slower work of diagnosing root causes. The same patterns hold true in American departments, where limited data fluency and a persistent gap between officers and analysts continue to hinder genuine problem-solving work.

The irony is that even in its shallow form, POP still produces measurable reductions. Both the Jersey City experiment and the Lowell, Massachusetts randomized control trial (which the National Academies acknowledged involved weak problem analysis) found the strategy effective, with Lowell producing a 19.8 percent reduction in calls for service despite officers implementing what researchers described as a general disorder policing strategy rather than carefully designed, problem-oriented responses. The ceiling for POP’s impact is considerably higher than what most agencies have achieved.

Intelligence-Led Policing: From Problems to Threats

If POP asks why crime persists, ILP asks who is responsible for the most serious harm. ILP originated in the United Kingdom in the early 1990s, when the Kent Constabulary used systematic analysis of offense patterns to identify prolific offenders responsible for a disproportionate share of property crime, producing a reported 24 percent crime reduction over three years. The Association of Chief Police Officers of England formalized the approach in 2000 via the National Intelligence Model. The concept crossed the Atlantic after Sept. 11, 2001, with the International Association of Chiefs of Police convening a Criminal Intelligence Sharing Conference in 2002 that led to a national intelligence architecture connecting local agencies to federal systems.

What distinguishes ILP from other data-driven approaches is the structured intelligence process behind it. Where precision policing is a deployment philosophy focusing officer time and enforcement on the small share of offenders and locations driving serious crime, ILP is an analytical engine producing the intelligence any precise approach depends on. It runs on a continuous cycle of planning collection requirements; gathering raw information from sources as varied as criminal records, informants, surveillance, and community tips; evaluating and analyzing that information; disseminating finished intelligence products to commanders; and reassessing whether those products actually informed decisions. The distinction between information and intelligence is what makes that process work—raw data only becomes intelligence after analysts vet it, evaluate it, and connect it across sources. That analytical step is what makes ILP distinct, not the targeting itself.

No scholar has done more to define ILP than Jerry Ratcliffe, a former British police officer and current professor at the University of Pennsylvania. In his foundational text Intelligence-Led Policing (now in its third edition), Ratcliffe argues that ILP is a “business model and managerial philosophy” in which crime intelligence drives objective decision-making, targeting prolific offenders, disrupting criminal networks, and allocating resources based on assessed risk rather than tradition or political pressure. His “3-i model”—interpret the criminal environment, influence decision-makers, and impact the criminal environment—provides the conceptual framework. ILP prioritizes repeat victims and prolific offenders, facilitating harm reduction through strategic management and targeted enforcement. As Ratcliffe himself has noted, the framework demands that analysts produce intelligence products that actually shape command decisions—a step many agencies struggle to institutionalize.

The Evidence Gap for ILP

While ILP has been widely adopted, the empirical base supporting its effectiveness remains thin. ILP’s most distinctive contribution is its focus on prolific offenders and high-harm individuals; however, a scoping review found that most experimental studies tested it in a narrower form, using crime intelligence to guide where and when officers were deployed. This closely resembles the hot spots policing approach examined in Part 3 of this series, making it difficult to isolate what ILP itself contributes beyond putting officers in high-crime areas. The 2018 National Academies report did not treat ILP as a standalone category, and Police Chief magazine has acknowledged the lack of empirical evidence in its favor. This does not mean ILP is ineffective; rather, it means the field has not yet subjected it to rigorous controlled evaluation. The gap is partly structural: ILP is a management framework, not a discrete tactical intervention that lends itself easily to randomized controlled trials. But it is also a challenge for policymakers being asked to invest in intelligence infrastructure on the strength of operational logic rather than controlled evidence. Centered on human analysts interpreting intelligence, ILP is conceptually distinct from predictive policing, which relies primarily on algorithmic models to forecast crime. Although the two approaches rest on different theories of how intelligence should drive deployment, they are increasingly used together. (Part 6 of this series will examine predictive policing in depth.)

Constitutional Boundaries and Accountability

ILP raises civil liberties concerns that are different in character from POP’s. Any approach built around watching specific people before they commit a specific crime creates real risks of profiling, over-surveillance, and mission creep—especially if the underlying intelligence is wrong, stale, or shaped by the biases of whoever collected it.

Agencies running federally funded intelligence operations must comply with 28 CFR Part 23, a federal regulation that sets the ground rules for what police can keep in a criminal intelligence database. It requires agencies to have a “reasonable suspicion” of criminal activity before retaining information about someone, limits how long that information can be held, and requires that unsubstantiated data be purged. It is the closest thing to a national standard for intelligence retention; however, compliance depends largely on each agency’s internal controls.

The track record on fusion centers—the regional intelligence-sharing hubs created after 9/11—illustrates what happens when those controls are weak. Originally, fusion centers were established to coordinate counterterrorism intelligence across federal, state, and local agencies. But their mission quickly broadened to an “all-crimes, all-hazards” approach that now covers everything from drug trafficking to cybercrime to disaster response. A 2012 bipartisan Senate investigation reviewed over a year of fusion center reporting funded as counterterrorism work and found the intelligence produced was “oftentimes shoddy, rarely timely, sometimes endangering citizens’ civil liberties,” with most reports having no connection to terrorism at all. Nearly a third were never released because they lacked useful information or potentially violated federal privacy protections. Some organizations have continued to document concerns about inadequate oversight and the retention of data on individuals never suspected of wrongdoing.

Agencies adopting ILP should build in data transparency, regular audits, written collection and retention policies, and clear lines of accountability for what analysts gather and how long it is kept. The reason these guardrails matter is that ILP’s targeting logic carries an inherent demographic risk. The 2018 National Academies report found that concentrating enforcement on high-risk people or high-risk places is likely to produce large racial disparities in police contacts—not because individual officers are biased in the moment, but because the underlying data used to identify “high-risk” individuals reflects decades of uneven enforcement. Closing that loop requires institutional design choices made before any officer-citizen encounter occurs: regular bias audits of intelligence products, documented criteria for who gets added to or removed from watch lists, civilian or independent oversight of analytic processes, and transparency in how targeting decisions are made.

The Adoption Challenge

Both POP and ILP demand sustained analytical reasoning about causes and actors, not just locations—a higher bar than many agencies can currently meet. Yet practical on-ramps exist:

• The Center for Problem-Oriented Policing at Arizona State University maintains an extensive library of problem-specific guides.
• The Herman Goldstein Award and the United Kingdom’s Tilley Award have generated more than 1,800 real-world submissions.
• The Bureau of Justice Assistance has published guides emphasizing that intelligence-led approaches can begin with structured patrol officer reporting shared through regional networks.

The analytical capacity these strategies require need not be built from scratch within every agency, as regional partnerships, shared analyst positions, and university collaborations can bridge the gap.

Assessment

POP benefits from a substantial evidence base that includes a 33.8 percent mean reduction in crime and disorder, a “Promising” CrimeSolutions rating, and a National Academies endorsement. It is inherently compatible with the resource constraints defining contemporary policing: Solving the underlying problem is more efficient than responding to the same incidents indefinitely. POP’s principal weakness is implementation fidelity. Agencies that invest in genuine analytical capacity and reward problem-solving outcomes—not just arrest totals—will realize far greater returns than those that adopt the language of POP without the discipline required.

ILP offers a theoretically compelling complement: a decision-making framework that prioritizes the highest-harm threats. Its weakness is empirical in that the controlled evaluations needed to establish its independent effectiveness do not yet exist. Agencies adopting ILP should do so with eyes open about both its promise and its risks, particularly regarding civil liberties and the potential for intelligence-driven enforcement to damage community trust in communities that can least afford it.

bool(true)

bool(true)

string(3) "100"

Next in this series: Part 5—Community-Oriented Policing: Building the Trust That Effective Policing Depends On