Every City Investigates Crashes. Very Few Actually Learn From Them.

That’s not an accusation. It’s a structural reality.

When a serious crash happens, there is a well-established response. Police arrive, a report is written, and all the details are recorded. That report is shared with the prosecutor and insurance companies and fault is assigned. In some cases, the crash becomes part of a larger pattern that is used as evidence that a particular intersection or corridor may need improvement. Maybe that helps justify funding for a signal, a turn lane, or another intervention.

From the outside, this looks like learning. A bad outcome occurs, information is gathered, and something changes.

But most of the time, that’s not what’s happening. 

What cities have built is a system for processing crashes, not for learning from them.

The Current Process is Designed to Answer a Very Specific Question

The traditional crash investigation model is designed to answer a very specific question: who is responsible?

That question shapes everything that follows.

Police reports focus on user behavior: who was speeding, who failed to yield, who was distracted. This information is essential for legal and insurance purposes. It helps assign liability and bring closure to individual cases.

But that same structure limits what the process can produce. If the goal is to determine fault, then the built environment becomes background context, not a subject of inquiry. The design of the street, the signals it sends to drivers, the cumulative effect of previous decisions. These environmental factors are rarely examined in a systematic way.

And just as important, the information often stops where it was collected.

The police department gathers information to reconstruct the actions that occurred minutes before the impact. This information is packaged into a report that is filed. And in many cities, it never becomes part of a shared, multidisciplinary conversation about what went wrong and what should change. 

At the same time the city engineer might see aggregate data over time of repeated crashes. Public Works might hear complaints from residents about high speed or near misses. Community members may have their own lived experience of the same location.

But those pieces are not routinely brought together around a single question: what did this crash reveal about the system?

So the same patterns repeat. The same locations show up again and again. And each individual crash is treated as a discrete event, rather than part of a larger failure.

Cities Are Becoming More Proactive

A few years ago, the city of Ann Arbor, Michigan, began to approach street safety and crashes differently. Like many cities, Ann Arbor adopted a Vision Zero plan with the goal of eliminating traffic fatalities and serious injuries. But they didn’t stop at just setting a goal. They built a practice around it.

As part of that effort, the city created a Crash Response Team: a standing, multidisciplinary group tasked with reviewing every serious crash. Not just documenting it, but reviewing each of these crashes to understand the contributing factors.

That distinction matters.

Instead of relying on a single report, the team brings together staff from different disciplines to look at each crash from multiple angles. They begin each review with the assumption there are multiple contributing factors. They examine not just user behavior, but the design of the street, the surrounding context, and the institutional decisions that shaped both.

The goal of this team is not to assign blame. It is to understand how the system behaved and where it failed.

When Ann Arbor invited Strong Towns to facilitate a Crash Analysis Studio workshop, it wasn’t because they lacked a process. It was because they wanted to strengthen one they had already built. That, more than any specific tool or methodology, is what sets them apart.

A Different Kind of Discipline 

What Ann Arbor demonstrates is not a new program, but a different kind of discipline.

They have made crash review a repeatable, institutional practice. It is something the city does consistently, with structure and intention. It brings together multiple perspectives. It assumes complexity. And it is explicitly oriented toward learning.

That’s a different posture than most cities take.

In many places, learning is incidental. It happens if someone notices a pattern, if a project happens to align with a known issue, or if a particularly severe crash draws attention. But it is not built into the routine operation of the system.

Ann Arbor has moved that learning into the center of their process.

They are not waiting for patterns to emerge over years. They are examining each serious crash as an opportunity to understand how their streets function in the real world. And they are doing it in a way that allows different kinds of expertise to inform each other.

This is not about having better data. Most cities already have the data they need. It is about creating the conditions where that information becomes shared understanding and where that understanding leads to change.

A Commitment to Practice

It’s tempting to look at a city like Ann Arbor and assume the difference is resources, staffing, or political alignment. Those things help. But they are not the core issue.

The core difference is a commitment to practice.

Crash investigation, in most cities, is a task that gets completed. In Ann Arbor, crash review is an ongoing activity that the institution has chosen to take seriously. It is structured, repeated, and continuously examined for improvement.

That shift from reacting to individual events to building a practice of learning is what turns a safety plan from a document into something real.

The City of Ann Arbor has created an interactive Traffic Crashes Dashboard for the public. The dashboard is showing a trend of a decline in the number of crashes, and keeps a focus on reaching zero fatalities. 

A Repeatable Approach

Other cities don’t need to replicate Ann Arbor’s exact structure to begin moving in this direction, but they do need to recognize what’s missing.

If crash data stays within a single department, if reports are written and filed without broader review, if the built environment is treated as fixed rather than examined then the system will continue to produce the same outcomes.

The first step is not a new tool or a new policy. It is a decision to treat crashes as something the city can learn from, together, on a regular basis.

That requires structure. It requires bringing the right people to the table. And it requires the discipline to keep doing it, even when the answers are uncomfortable or incomplete.

Ann Arbor’s example shows that this is possible. Not because they have solved the problem, but because they have committed to learning from it.