Road safety data is most useful when it helps answer a decision: where should a city investigate, which change should it try, and how will it know whether conditions improved? A map of reported crashes can identify places to study, but it does not describe every near miss, explain why a crash happened, or show the full experience of people using a street.

Combine sources, understand their limits

Crash records describe events reported to authorities and include the details those systems collect. Their coverage and level of detail can vary. A street with few reported crashes may still have high speeds, difficult crossings, or many close calls. Conversely, a cluster of records needs context before it can explain what design or operating change is appropriate.

Agencies can pair crash data with speed and traffic observations, roadway characteristics, transit and walking conditions, maintenance reports, and input from people who use the location. Record how each data source was collected and what it leaves out. Take care with small counts: one additional event can substantially change a percentage when the underlying number is low.

Choose measures that match the goal

The Federal Highway Administration’s Safe System approach describes a framework that accounts for human mistakes and the vulnerability of the human body, with responsibility shared across the transportation system. For a local project, that means stating the safety problem and desired outcome before selecting a treatment or performance measure.

Measures might include observed operating speeds, crossing opportunities, yielding behavior, or serious injury outcomes, depending on the site and intervention. Use measures that can be collected consistently, and be clear about whether a result is an observed change, an estimate, or a long-term outcome that cannot yet be assessed.

Evaluate the change, not just the dashboard

Set a baseline and an evaluation period before construction or operation changes. Track whether the intervention was installed as intended, then compare relevant conditions over time. Seasonal travel, nearby construction, enforcement changes, and traffic diverted to adjacent streets can all affect what a before-and-after comparison appears to show.

Share methods and limitations alongside results. If a project changes conditions on one corridor, examine nearby streets so a local improvement is not mistaken for a network-wide benefit. Involve community members in interpreting results, especially where formal datasets do not capture daily barriers or near misses.

Turn evidence into a repeatable practice

A strong safety program connects data to a decision, gives the public a way to understand that decision, and returns to measure what happened. It also treats privacy responsibly: use information at the level needed for analysis and avoid exposing identifiable trip histories.

Data does not replace engineering judgment or lived experience. It can help cities make priorities visible, test whether an action had the intended effect, and identify where more investigation is needed. For another perspective on street operations and evaluation, read The connected street is a better street.

Sources