Seattle and other cities use predictive data to allocate curb space for maximum effect.
By Cole Jaillet
Cities across North America are actively working to adapt to new curbside activities while ensuring safe and equitable access for their citizens. Significant progress has been made, but it has come with challenges. One of the more subtle yet considerable contributors to these challenges has been the traditional methods used to manage curb space, such as short-term and long-term parking and designated parking zones. Although these uses remain relevant, curbside activity has expanded to include rideshare services, food pick-up and delivery, micro-mobility options, electric vehicle (EV) charging stations, and parklets.
Curb management has emerged in recent years as an answer to these challenges. With the evolution of traditional parking management, curb management aims to adapt these long-standing methods to meet the new and ever-evolving demands of constituents. Although curb management has informed better decision-making and policy change, it has required a sound technical foundation capable of accommodating frequent change and continued insight.
The partnership between Seattle's Department of Transportation (SDOT) and Turnstone Data Inc., a software platform that provides comprehensive curb occupancy data by predicting unpaid parking behavior, exemplifies how the public and private sectors can collaborate to position technological innovation at the core of parking programs to support evolving curb demands.
A brief history of Seattle's rate program
In 2010, SDOT embarked on an ambitious initiative to manage parking rates using routinely collected data. The policy, established by elected officials in Seattle, aims to ensure that one to two parking spaces per block face are open and available throughout the day. This policy was grounded in several key principles:
• Review occupancy regularly, by neighborhood, for each hour of the day.
• Analyze results for peak periods within the time-of-day rate bands (defined as 9 to 11 a.m., 11 a.m. to 5 p.m., and 5 to 7 p.m.).
• When analyzed, if occupancy is greater than 85%, raise rates by $0.50 per hour.
• When analyzed, if occupancy is lower than 70%, lower rates by $0.50 per hour.
Initially, this performance-based model relied heavily on yearly manual data collection. SDOT collected data for one day in each parking area in the spring and changed the parking rates based on the metrics in the fall. This labor-intensive process limited the frequency of rate adjustments, constraining the program's responsiveness to seasonal demand fluctuations.
The catalyst for change
2020 presented unprecedented challenges for urban mobility, especially in downtown areas. To address the effects of the pandemic, SDOT set all metered parking rates to free parking for 3 months city-wide.
By July 2020, as cities began navigating the complexities of reopening, SDOT reinstated parking fees at a flat rate of $0.50 per hour across all neighborhoods. This flat rate departed from the pre-pandemic pricing model and did not reflect the varying demand across different areas and times. The limitations of the existing system became apparent: returning to pre-pandemic operations using the traditional approach would be a lengthy process, likely spanning a decade due to the incremental nature of annual rate adjustments.
Recognizing the urgency for modernization, SDOT sought a solution to overcome these operational hurdles. The department’s partnership with Turnstone emerged as a strategic relationship that leveraged technology to enhance efficiency, responsiveness, and data-driven decision-making in parking management.
Innovating occupancy analysis
To address the challenges, SDOT and Turnstone collaboratively introduced a predictive modeling platform to comprehensively understand continuous, year-round parking demand without relying on extensive hardware installations like cameras or sensors. The approach encompassed several methodologies:
• Inventory the supply of spaces per block face and incorporate temporal adjustments such as reservations, construction permits, and hourly usage changes. Then, summarize this block-face-level information at the neighborhood and city levels.
• Monitor reservation feeds and meter removals to track changes in supply over time.
• Combine transaction data from all hardware meters and mobile payment applications to determine the average number of paid spaces per hour on each block face, providing a baseline for understanding parking demand.
• Train a statistical curve from manually collected sample data that predicts the probability that a paid vehicle is still present after payment. Then, apply this calculation to paid parking to better understand all paid usage on a block face in a given hour.
• Extend the statistical model to forecast unpaid parking behavior. By categorizing usage into “stop-and-shop" (0 to 1 hour), “longer-stay” (1 to 4 hours), and “workday parkers” (4+ hours), the model provided insights into the extent of unpaid parking and its effect on overall occupancy.
• Feed the model with the starting conditions for a given day to account for hours when vehicles can park freely outside paid hours, so that it can predict parking patterns moving forward.
• Incorporate automated monitoring to understand future changes in parking behavior. This last approach was necessary because the learning model was trained on manually generated inputs linked to a single point in time. Knowing when the model was potentially becoming outdated would enable additional training.
• Conduct regular paid parking studies in each area. SDOT collects parking occupancy in each paid area at least once per year and for several areas multiple times per year to ensure it understands changes in parking behavior.
Transforming rate reviews
Deploying Turnstone's predictive modeling platform marked a significant turning point for SDOT as the program shifted from manual to statistical methods. The most immediate benefit was the ability to conduct rate reviews more frequently and efficiently. Freed from solely relying on manual data collection, SDOT increased the frequency of rate adjustments from once to 3 times per year. This agility enabled SDOT to adequately respond to seasonal changes by adjusting parking rates to maintain desired occupancy levels.
SDOT's new iterative approach to rate adjustments has had a noticeable impact on parking availability. Starting from the low baseline rate of $0.50 per hour, SDOT strategically adjusted rates based on the statistical model's insights. In downtown Seattle, occupancy levels decreased from a consistent 100% to below 85% during peak times. Even more useful for parkers, this occupancy level was equivalent to the 70th percentile, meaning one or more spots were available on most blocks for 7 out of 10 trips. This outcome marked a significant improvement from the previous 3 out of 10 that parkers had been experiencing.
Beyond improving occupancy rates and driver satisfaction, the enhanced rate review process yielded tangible policy benefits. SDOT staff regularly reviews the rate changes with the group of neighborhood business district leaders around the city known as the Curbspace Access Sounding Board. This group has been engaged to help shape program metrics and data collection methods. This way, the community discussion is less about whether the rate is, say, $3.50 per hour and more about the occupancy levels and what other curb management projects might help to support the business districts. The Seattle program is an iterative growing effort as more curb management and machine learning tools become available.
Occupancy data's versatility
The success of SDOT's partnership with Turnstone underscores the value of comprehensive curb data. However, not all cities are in a position to invest in the transition to demand-based pricing. Program maturity, technological infrastructure, and stakeholder readiness are among a few factors determining the appropriate starting point. The following are just a few alternatives available to support current and future goals:
• Enhanced inventory insights. Identify what parking assets exist, where they are located, and how they are utilized to inform strategic planning.
• Stakeholder engagement. Effectively communicate with businesses, residents, and policymakers using data-driven narratives that foster support for initiatives and investments.
• Public accessibility. Share real-time parking availability with the public through apps and online platforms to reduce congestion and improve the overall curb experience.
• Enforcement optimization. Prioritize areas of non-compliance when enforcing preferred behaviors to ensure higher compliance while optimizing resource use.
Data-driven exploration
Cities across North America are demonstrating how parking data can be leveraged to influence some of these alternative desired outcomes without making immediate large-scale changes to pricing structures.
For example, the City of Boston sought to understand the potential of parking meter transaction data as a proxy for physical occupancy data. In partnership with Turnstone, the city expanded its 2017 performance parking pilot to focus on the Back Bay and Beacon Hill neighborhoods, both of which are known for high congestion. By implementing continuous, year-round automated data collection, Boston achieved approximately 89% accuracy in occupancy predictions with minimal model refinement. The software-centric approach resulted in significant cost savings, reducing daily data collection expenses from approximately $100 per block face to just $1, enabling the city to take a scalable, data-centric approach to policy discussions.
Similarly, the City and County of Denver partnered with Turnstone to better understand parking demand throughout the city after making its first on-street paid parking rate change in 20 years. Through this partnership, Denver successfully upheld its initial price change — a challenging accomplishment — and is now considering implementing its first performance pricing program in Cherry Creek, Denver’s premier mixed-use neighborhood, featuring fashion boutiques, art galleries, and more than 350 stores and restaurants.
By accessing a city-wide view of complete parking demand, cities like Boston and Denver have been able to continuously inform stakeholders of changes in parking demand, the factors that influence change in demand, and the best ways to allocate resources to optimally support their constituents’ needs, all while building a foundation for future policy development.
Rightsizing the data for your program
The journey toward intelligent curb management is not a one-size-fits-all endeavor. Each city must navigate its unique challenges, opportunities, and stakeholder priorities. The experiences of Seattle, Boston, and Denver illustrate that the key to progress is embracing data-driven solutions that can adapt and grow over time.
For cities contemplating their next steps, several considerations can guide the process:
• Assess program maturity. Understanding the current state of parking management systems and technological capabilities is crucial. This assessment informs the feasibility of implementing advanced solutions like predictive modeling.
• Engage stakeholders early. Collaborating with businesses, residents, and policymakers ensures that initiatives align with community needs and garner necessary support.
• Prioritize scalable solutions. Selecting technologies and methodologies that can evolve with the program facilitates gradual implementation and adaptation to changing circumstances.
• Measure and communicate impact. Regularly evaluating the outcomes of new strategies and sharing successes and challenges build trust and transparency with the public.
• Curb management challenges are complex and multifaceted, but they also present opportunities for innovation and improvement. By harnessing the power of predictive modeling and comprehensive data analysis, cities can transform how they manage curb spaces, enhancing efficiency, increasing funds, and improving the user experience.
The journey is ongoing, and as more municipalities adopt intelligent parking solutions, collective knowledge and best practices will continue to evolve. What's clear is that the future of mobility depends on our ability to adapt and innovate, leveraging technology to create a more dynamic curb.
COLE JAILLET is the executive vice president of growth for Turnstone. He can be reached at cole@turnstonedata.com.