Best practices for new intervention rules

Many cities are adopting or considering 30 km/h urban speed limits.
For an administration, the political decision is only half the battle: the other half depends on the credibility of the limit.
If actual speeds remain high, the measure causes frustration and conflict, and becomes difficult to defend over time.

The gap between speed limits and actual speeds: what the data from Milan shows
A study led by the Senseable City Lab at the Massachusetts Institute of Technology (MIT) offers a snapshot that is rare in terms of its size and detail: 50.98 million speed observations collected in Milan in January, April, July, and October 2023.
The authors analyze the entire road network and use, among other key indicators, the 85th percentile speed, often used in traffic engineering to describe the behavior of the “fastest but most common” drivers.

The most significant data for decision-makers indicates that on roads with a 30 km/h speed limit, the average speed is 26.98 km/h, while the 85th percentile reaches 38.26 km/h.
Non-compliance is particularly evident at night, with maximum speeds between 2:00 and 3:00 a.m., when traffic volumes drop.
In operational terms, this means that the same stretch of road may appear to be ‘under control’ during the day, but transform into a completely different scenario at night, when social control and congestion are significantly reduced.

Two micro case studies from 2023 make the message even more concrete.
In the Porta Volta and Isola areas, where the Municipality of Milan reduced the speed limit from 50 to 30 km/h, the change did not lead to a substantial decrease.
In Porta Volta, the average speed fell from 23.33 to 22.70 km/h. In Isola, it rose from 25.48 to 25.70 km/h. And the percentage of vehicles exceeding 30 km/h increased in both cases: from 12.19% to 13.29% in Porta Volta and from 13.02% to 13.29% in Isola.
The statistical checks reported in the study indicate that, in these two contexts, the reduction in the speed limit did not produce a significant change.

The picture does not change when the analysis is broadened.
Comparing 1,694 similar road segments, some with a speed limit of 30 km/h and others with a speed limit of 50 km/h, the estimated causal effect of the lower speed limit is modest: 2.29 km/h less on average speed and 3.45 km/h less on the 85th percentile.

The authors’ conclusion is clear: the introduction of the lower limit alone is not sufficient to effectively reduce speeds.

Street design and priorities: where 30 km/h can really work
So where does the game play out? In the signals that the road sends to drivers.
What is noticeable is the association that sees lower speeds as a road network built on a road network with tighter connections and denser residential buildings.
Conversely, longer segments, more lanes, and a more “open” field of vision are linked to higher speeds.
The presence of bus stops is also correlated with lower compliance, consistent with the idea that they are often located on wider, faster roads, where the design encourages acceleration.

For a public administration, the most useful step is to translate this into priorities.
In practice, this has translated into applying a model that prohibits speeds exceeding 30 km/h to 1,959 km of roads in Milan, currently limited to 50 km/h, excluding primary and secondary roads.
The result is a segmentation that can guide choices and budgets: 335 km with 85th percentile predicted to be below 30 km/h; 796 km between 30 and 40 km/h, where the limit would require additional measures; 828 km above 40 km/h, where a ‘30 km/h zone’ is likely to fail without major changes or enforcement.

The lesson for those in government is pragmatic: before extending a limit, it is worth understanding whether the road already “speaks” at 30 km/h or whether it communicates something else entirely.
Only in the latter case will the choice not be between implementing or not implementing Zone 30, but between creating it truly and concretely, with interventions consistent with the context, or accepting that it may remain a number on a sign.

Source: Orsi G., Venverloo T., La Grotteria A., Fugiglando U., Duarte F., Santi P., Ratti C. (2025) “Street design and driving behavior: evidence from a large-scale study in Milan, Amsterdam, and Dubai”, Senseable City Lab – Massachusetts Institute of Technology