Slow Streets Performance Dashboard

Track the performance of San Francisco’s Slow Streets, powered by Telraam

Slow Street in San Francisco

How are San Francisco’s Slow Streets performing against safety targets?

The San Francisco Municipal Transportation Agency established speed and vehicle volume targets for Slow Streets in December 2022. Slow Streets have a target speed of 15mph and target vehicle volume of 1000 cars per day. With our network of Telraam sensors, we are tracking performance of 8 of the 16 slow streets against these targets.

Key findings

Vehicle Volumes

Most Slow Streets with sensors meet the City’s vehicle volume performance goal. There are two major exceptions: On most weekdays, Lake Street exceeds the vehicle volume target. Occasionally Page Street has also exceeded 1,000 vehicles per day.

Speed

Two Slow Streets are persistently – and dangerously – out of compliance with the speed performance goal: Hearst Avenue and Cayuga Street. Most Slow Streets are also out of compliance with the speed performance goal at least some of the time.

Of the eight Slow Streets where there are live Telraam S2 Sensors, only Sanchez Street and Golden Gate Avenue consistently report average speeds below the 15 mph target (more on both below).

Explore the data

KidSafe SF operates a network of Telraam S2 traffic sensors on Slow Streets and across San Francisco. These sensors track speeds and vehicle volumes on 8 of the 16 Slow Streets. This dashboard shows performance against the two key metrics established by the SFMTA board: vehicle volumes and target speed.

Recommendations

When the Slow Streets Program was made permanent, the SFMTA Board empowered the staff to calm the Slow Streets by implementing elements from the 2023 Slow Streets Design Toolkit – median diverters, speed cushions, roadway narrowing – without SFMTA Board for approval for each new design. The intent of this program is to allow the staff to iteratively achieve the outcomes of slower, people-friendly streets on the SFMTA Board-approved Slow Streets. 

However, SFMTA has delivered virtually none of the improvements to address the dangerous and persistent speeding conditions on our Slow Streets. To improve Slow Streets to be a safe, inviting experience for all road users, SFMTA should do the following.

Implement traffic calming measures on Slow Streets that are out of compliance

Median diverters, speed cushions, and roadway narrowing should be installed on streets that are persistently out of compliance with the program.

Every intersection between a Slow Street and an arterial road should have a median diverter or concrete island modal filter to prevent speeding, cut-through traffic, while also allowing residents to reach their homes.

Roadway narrowing, both with mid-block pinchpoints and at intersections with curb bulbouts, is an effective and proven technique to slow speeds.

These treatments are already part of SFMTA’s Slow Streets design toolkit. SFMTA should move quickly to use the existing toolkit to calm traffic on our Slow Streets, and make these streets successful.

Pinchpoint on a slow street in the Presidio in San Francisco

Concrete island diverter in Sacramento

Median diverter at Page and Octavia St in San Francisco.

A note about the data collected by Telraam sensors

SFMTA also evaluated the performance of all Slow Streets once since the Program was made permanent. In April 2023, SFMTA published its Slow Streets Evaluation Report, which similarly found that most streets meet the vehicle volume performance goal, but do not meet the speed goal.

Some of our Telraam measurements differ from what SFMTA found in its April 2023 Slow Streets evaluation report. We offer a few explanations:

  1. Point in time versus constant data: To do its Slow Streets evaluation, SFMTA hired a contractor to lay pneumatic tubes for 48-hour periods on several points along individual Slow Streets. As SFMTA shared in the Slow Streets Evaluation Report, most of that data collection happened in January 2023. While the method of pneumatic tubes is an accurate way to gather data on vehicle behavior for a single point in time, it misses significant changes over the course of the year. For example the traffic patterns on Hearst Avenue change in the rainy season, and they varied with both the start and stop of the academic calendar due to the two elementary schools on the street. By contrast, Telraam data is continuous (while sunlight is available) so trends on a street overtime are recorded.

  2. Block conditions vary considerably along a corridor: The speed and vehicle volume profiles of a block can differ on the same Slow Street depending on the land use of the street. For example, the Sanchez Street Telraam S2 sensor is located on a block with a church and a small parking lot. Due to high levels of people walking on the street combined with the narrow entrance to the parking lot, the drivers who use that block of Sanchez typically drive very slow (~5 mph), which KidSafe verified with a speed gun in September 2023. Given the unique conditions of this block, more sensors along Sanchez Street would likely show higher average vehicle speeds on the Slow Street.

  3. Intersections distort speed: The sensors closest to intersections, like the one on Golden Gate Avenue, underreport the speeds of drivers because of the acceleration and deceleration required to navigate an intersection. More sensors along Golden Gate Avenue would likely show higher average vehicle speeds.

  4. Sensor speed measurements are estimates: Telraam S2 sensors can estimate vehicle speeds with a margin of error of 10%. This capability means we can be confident in the Telraam-reported speed trends directionally, but the sensor doesn’t have the same degree of accuracy as a speed gun or pneumatic tubes. With Telraam, we’ve shared some video recordings of driver behavior from Hearst Avenue to ensure that the Telraam S2 sensor is calculating vehicle speed as accurately as possible, especially at higher recorded speeds.

  5. Telraam will record fewer vehicles as days shorten: Because the Telraam S2 sensors rely on daylight to record traffic volumes and speeds, as the days shorten the overall counts will decrease. For example, on September 26, 2023, the sun rose at 7am PDT and set at 7:00pm PDT for 12 hours of daylight. However, on November 25, 2023, the sun rose at 7:00am PST and set at 4:52pm PST for just under 10 hours of daylight. Given that common traffic patterns like commuting don’t change much with respect to sunlight, it is safe to assume people are still traveling on the street in the dark for those two hours, but due to the lack of light the Telraam S2 sensors won’t count them. 

  6. Sensor replacements: Over the course of this project, we’ve had to replace a few sensors due to hardware issues. When we’ve observed issues with the sensors – constant rebooting, inaccurate counts – we’ve reported them to Telraam for replacement. Thus there have been some changes in the data, which reflect the replacement of a more accurate sensor. For example one of the S2 sensors on Slow Lake needed to be replaced in mid-September 2023, which led to an increase in reported traffic volume across all four road user categories.

Telraam S2 sensor in a window

KidSafeSF + Telraam

Learn more about our partnership with Telraam, and how you can collect traffic data to help us hold SFMTA accountable to street safety outcomes in San Francisco.

Make a donation.

Support KidSafe SF’s advocacy work to make San Francisco a safer, greener city.