California Open Transit Signal Priority (TSP) Performance Metrics
Version 1.0 – Finalized April, 2026
Purpose
The purpose of the California Open Transit Signal Priority Performance Metrics is to provide definitions and data requirements for a uniform set of performance metrics that can evaluate Transit Signal Priority (TSP) performance across the state. The metrics are intended to help Caltrans and partner agencies assess the effectiveness of TSP systems at an intersection and corridor level on an ongoing, operational basis. In addition, they are intended to promote uniform performance analysis and open data standards in this area.
TSP is a technology that alters the phase of traffic signals to allow transit vehicles to traverse intersections faster. Without TSP, traffic signals may operate dynamically or statically, but do not specifically work to prioritize transit vehicles. When signals operate with “pre-timed” static operations, each signal phase operates for a set amount of time in a set sequence. With dynamic or “actuated” signal control, signals can change their timings based on the input of sensors to achieve outcomes such as greater vehicle capacity. Applying TSP can encompass a variety of strategies in parallel to these measures, such as extending a green light for a longer time or ending a red light earlier to accommodate a transit vehicle from an opposing direction. In some systems, more advanced strategies are used, such as providing transit-only phases or dynamically adjusting signal timing along a corridor to prioritize transit vehicles.
In the US, most TSP systems are either localized “hardware-based” systems or centralized “cloud-based” systems. In the latter, vehicles provide their locations to a centralized system that then makes requests to Traffic Signal Controllers, whereas in the former, vehicles make requests directly to controllers by activating a detection system. TSP systems can also be supported by other engineering interventions, such as transit-only lanes, stop spacing, stop placement, and intersection redesign (which can reduce the length of signal phases or restrict movements that conflict with transit). Regardless of their mechanism, the goal of these TSP strategies is to create a street network that is more supportive of transit.
Jurisdictions implementing TSP do so for different reasons. For example, some systems are created with the goal of increasing general transit running speeds, whereas others are primarily intended to support better transit schedule recovery. Different jurisdictions will have different levels of tolerance for impacts to general traffic. To account for these different needs, TSP systems implement varying Business Rules and operating parameters to govern how they function. For instance, if a jurisdiction has the goal of improving transit reliability but has little tolerance for delays to general traffic, they may configure the system to only grant priority to late vehicles. Accurately capturing the impact of TSP across these differing goals means a variety of metrics are necessary to properly measure performance across jurisdictions.
These nuances also make performance measurement a critical challenge to implementing TSP. For a 2020 Transit Cooperative Research Project report, researchers surveyed agencies and found that “Measuring the TSP system’s benefit” was the greatest perceived challenge in operating such a system. This reflects a challenge in integrating data from many different components, isolating the impact of TSP in a complex operating environment, and balancing the different priorities that different performance metrics can reflect. Transit agencies interviewed by Caltrans also found the lack of standardized methods for such analyses to be a significant challenge, hence the need for more open performance metrics in this area.
In order to create standardized metrics for different types of TSP systems, the metrics presented in this document focus on high-level service outcomes. While there are many other factors that are important to TSP implementation, such as lifecycle costs, these are not considered to be in scope for this document. Where possible, the metrics are intended to be implemented with standardized and publicly available data. Therefore, they focus primarily on results that impact transit service, avoiding factors that may vary between different TSP systems.
These metrics are intended for use on local streets and conventional highways with signalized intersections. While many of these metrics may be broadly applicable to other contexts, such as ramps and non-signalized highways, additional tools are likely necessary to measure transit performance in this context.
The Caltrans Division of Data & Digital Services (Caltrans DDS) intends to implement these metrics on a set of demonstration corridors where the necessary data is available in 2026. Based on the success of this demonstration, these metrics are intended to be refined over time and implemented for a larger set of corridors to measure the performance of TSP programs across the state. Caltrans DDS will release all software and documentation relating to these metrics under an Open-Source License, to allow for the broader implementation of TSP performance analysis.
Certain metrics are designated as Primary Metrics, whereas the others are designated as Additional Metrics. Primary Metrics are intended to measure the performance of systems for comparison across the state. Additional Metrics are intended to provide context for these Primary Metrics but are expected to be influenced by a greater number of factors and are therefore less appropriate for directly comparing performance across systems.
These metrics were shaped by the following principles:
- Measure whether TSP systems have met the goals of improving bus speeds and reducing the operating expense per Vehicle Revenue Hour.
- Operationalize metrics that are in use by other agencies and/or supported by academic literature.
- Emphasize metrics that can be operationalized using standardized data such as General Transit Feed Specification - Realtime (GTFS Realtime), but that support expansion using agency-specific datasets where applicable.
- Provide actionable insights regarding TSP performance as it impacts transit operations and service goals, such as ability to operate service at a higher frequency with a similar number of vehicle revenue hours.
These metrics were grouped into the following four categories, to measure different ways in which TSP can affect the transportation system. Two such metrics were designated as Primary Metrics (shown in bold text). All other metrics are designated as Additional Metrics.
-
Speed Metrics
- Trip Duration
- Trip Segment Speed
-
Granular Trip Time Metrics
- Dwell Time
- Intersection Delay (Transit)
- Moving Time
-
Variability Metrics
- Trip Duration Variability
- Trip Speed Variability
- Schedule Adherence at Timepoint
- Average Deviation from Schedule
- % Time with Excess Wait
-
Non-Transit Vehicle Metrics
- Non-Transit Vehicle Intersection Delay
- Non-Transit Vehicle Segment Travel Time (On-corridor)
- Non-Transit Vehicle Segment Travel Time (Cross-Street)
Speed Metrics
These metrics measure the overall speed and travel time of buses. This outcome is highly dependent on factors besides TSP effectiveness, such as general traffic conditions, other implemented priority measures, and scheduling decisions. However, it is important to report since it most directly reflects the cost of providing transit service and the passenger experience. These metrics can be contextualized with additional operational information, such as the travel times needed to decrease Scheduled Headways without increasing revenue hours.
The two Speed Metrics – Trip Duration and Trip Segment Speed – are defined in table 1:
Table 1
| Trip Duration | Trip Segment Speed | |
|---|---|---|
| Purpose | To measure the overall impact that TSP and other components of the transportation system have on the usefulness and cost of operation of transit service. Additionally, to inform changes to transit frequency enabled by TSP. | To identify areas where the operational cost and usefulness of transit service has changed before and after TSP installation. |
| Definition | The trip time for trips from the start of the TSP corridor to the end of the TSP corridor, reported as 50th and 80th percentile for each route along the corridor. | The speed as estimated for stop-to-stop segments through the Caltrans Speedmaps Tool, reported as 50th and 80th percentile for each route along the corridor. |
| Necessary Data | GTFS Realtime data from Caltrans Data Warehouse. 1 | GTFS Realtime data from Caltrans Data Warehouse. |
| Reporting Grain | Per Service Pattern, Day Part and Reporting Period. | Per stop-to-stop segment, Service Pattern, Day Part, and Reporting Period. |
| Potential Refinements | Report travel times for individual trips, rather than grouped by Reporting Period. | Report speeds for smaller segments by using higher-frequency Computer Aided Dispatching / Automated Vehicle Location (CAD/AVL) data, rather than less frequent GTFS Realtime data. |
| Desired Data for Refinements | A mapping of unstable GTFS Realtime Trip IDs to a stable Trip ID, to allow the comparison of trips over a longer period and inform service decisions. | High frequency CAD/AVL data. |
| Considerations | A decrease in trip duration is associated with lower costs to operate and improvements in accessibility. Specific thresholds for trip duration are associated with positive outcomes such as being able to reduce the number of vehicles needed to operate revenue service and being able to provide an improved transit frequency. For this reason, transit providers can use this metric to set goals for TSP outcomes. However, it is affected by many different factors, including scheduling decisions, and should be used in the context of other metrics. | Segment speed can be used to provide a visualization of where TSP has an impact on changing travel times, and to show where other transit priority interventions should be considered. Like travel times, it is affected by many different factors, including scheduling decisions. |
Granular Trip Time Metrics
These metrics quantify the amount of time that transit vehicles spend in different operating conditions, i.e. dwelling at stops, at intersections, and while moving. Measuring these parts of a transit vehicle’s trip can show the impact of interventions like TSP to be separated from scheduling and non-signal delay. However, they depend on access to more granular data, such as high-frequency CAD/AVL data and door status. This means that initially, they may only be implemented for certain systems where such data is readily available.
The three Granular Trip Time Metrics – Dwell Time, Intersection Delay (Transit), and Moving Time – are defined in table 2. Intersection Delay (Transit) is designated as a Primary Metric.
Table 2
| Dwell Time | Intersection Delay (Transit) | Moving Time | |
|---|---|---|---|
| Purpose | To measure the time that transit vehicles spend loading and unloading passengers. | To measure the time that transit vehicles spend at red signals. | To measure the time that transit vehicles spend moving. |
| Definition | The amount of time that a bus spends between first opening its doors at a stop and closing its doors for the last time at a stop. Reported as 50th Percentile and 80th Percentile. | The amount of time that buses spend stopped without their doors open around signals. Reported as 50th Percentile and 80th Percentile. | The total trip time with intersection delay and dwell time subtracted. Reported as 50th Percentile and 80th Percentile. |
| Necessary Data | High frequency CAD/AVL data with door open/closed data available. | High frequency CAD/AVL data. Door/open close data from CAD/AVL if near side stops are to be considered. | High frequency CAD/AVL data with door open/closed data available. |
| Reporting Grain | Per Service Pattern, Day Part, and Reporting Period. | Per Service Pattern, intersection, Day Part, and Reporting Period. | Per Service Pattern, Day Part, and Reporting period. |
| Potential Refinements | NA | Only count intersection delay while the signal is red, to better capture the impact of the closest signal to the bus compared to that of general traffic flows over the entire corridor. | NA |
| Desired Data for Refinements | NA | Traffic Signal Controller status over time. | NA |
| Considerations | A decrease in dwell time could contextualize changes in moving time, since dwells are more heavily affected by factors such as ramp usage, schedule padding, and ridership. | A decrease in Intersection Delay could suggest that buses are able to traverse signals more quickly. Since TSP affects signals at intersections, this would suggest that the system is having an impact on service. | An increase in moving time indicates that buses are able to move faster along the route. This could provide context for changes in intersection delay and dwell time. |
Reliability Metrics
On a statewide level, signal delay is a substantial cause of delays in transit service. As a result, improved service reliability is an important expected outcome of TSP. By increasing reliability, TSP can allow systems to decrease schedule padding and operate service more quickly and efficiently. Changes in service reliability also allow for evaluation of TSP before schedule changes are made, making it especially important for informing service planning.
The five Reliability Metrics – Trip Duration Variability (Table 3), Trip Speed Variability (Table 3), Schedule Adherence at Timepoint (Table 4), Average Deviation from Schedule (Table 4), and % Time with Excess Wait (Table 5) – are defined in the tables below. Trip Duration Variability is designated as a Primary Metric.
Table 3
| Trip Duration Variability | Trip Speed Variability | |
|---|---|---|
| Purpose | To determine whether TSP is having an impact on reducing overall travel time variability. | To determine the areas in which TSP is having an impact on reducing travel time variability. |
| Definition | The 80th/20th Percentile ratio of durations between specified points on a vehicle’s trip. Reported for the start of the TSP corridor to the end of the TSP corridor and between each Timepoint along the TSP corridor. | The 80th/20th Percentile ratio of trip speeds as measured for stop-to-stop segments through the Caltrans Speedmaps Tool, reported for each route along the corridor. |
| Necessary Data | GTFS Realtime data from Caltrans Data Warehouse and Timepoint locations. | GTFS Realtime data from Caltrans Data Warehouse. |
| Reporting Grain | Per Timepoint, Service Pattern, Day Part, and Reporting Period. | Per stop-to-stop segment, Service Pattern, Day Part, and Reporting Period. |
| Refinement Potential | Report travel times for individual trips, rather than grouped for Reporting Period. This could allow the comparison of trips over a longer period and inform service decisions. | Report speed variability for smaller segments by using higher-frequency CAD/AVL data, rather than less frequent GTFS Realtime data. |
| Desired Data for Refinements | A mapping of GTFS Trip IDs to stable Trip IDs. | High frequency CAD/AVL data. |
| Considerations | Trip duration variability decreasing could indicate that TSP is reducing long signal delays and/or that a system which treats late buses differently from other buses is having an impact. | Speed variability can be used to provide a visualization of where TSP has the most impact on variability. However, using GTFS Realtime limits granularity of the segments that can be displayed. |
Table 4
| Schedule Adherence at Timepoint | Average Deviation from Schedule | |
|---|---|---|
| Purpose | To determine whether TSP is causing transit vehicles to better adhere to schedules. | To determine whether TSP is causing transit vehicles to better adhere to schedules. |
| Definition | At each transit stop, the proportion of trips that are less than five minutes late and less than one minute early. | The difference between when the bus is scheduled to arrive at each stop and when the bus arrives based on real time data. Reported as 50th and 80th percentile. |
| Necessary Data | GTFS Realtime data from Caltrans Data Warehouse. | GTFS Realtime data from Caltrans Data Warehouse. |
| Reporting Grain | Reporting Period and Day Part. | Reporting Period and Day Part. |
| Iteration Potential | NA | NA |
| Desired Data | NA | NA |
| Considerations | An increase in schedule adherence could indicate that TSP is effective at increasing service reliability. In the absence of schedule changes, it can also indicate that buses are more easily able to recover from delays, indicating that TSP is having an impact on service. | An increase in average deviation from schedule could show that buses are arriving early more frequently and arriving late less frequently, indicating that TSP is having an impact on service. |
Table 5
| % Time with Excess Wait | |
|---|---|
| Purpose | To identify the impact of service reliability on the service experienced by passengers. |
| Definition | The proportion of the time during the Day Part in which an arriving passenger would experience an Actual Headway that is greater than the Scheduled Headway. |
| Necessary Data | GTFS Realtime data from Caltrans Data Warehouse. |
| Reporting Grain | Service Pattern, stop, Day Part, and Reporting Period. |
| Potential Refinements | Provide this data at a greater level of precision by using higher frequency data. |
| Desired Data for Refinements | High frequency CAD/AVL data. |
| Considerations | A decrease in % Time with Excess Wait could indicate that a TSP system is helping buses better match service expectations. This metric is useful to connect the service delivered by transit agencies to passenger experience and can provide additional context for schedule-based metrics. |
Non-Transit Vehicle Metrics
By altering signal timings and intersection layout, TSP may have an impact on travel time for non-transit vehicles. It is important to measure these impacts to non-transit vehicle travel time to determine the full impact of TSP. These metrics may be more difficult to calculate, since they depend on data that is less commonly made public. Future development of this document will consider incorporating new data sources to measure larger scale cross-street and corridor-wide traffic impacts.
The three Non-Transit Vehicle Metrics – Non-Transit Vehicle Intersection Delay (Table 6), Non-Transit Vehicle Segment Travel Time (On-corridor) (Table 7), and Non-Transit Vehicle Segment Travel Time (Cross-Street) (Table 7) – are defined in the below tables:
Table 6
| Non-Transit Vehicle Intersection Delay | |
|---|---|
| Purpose | To measure the impact TSP has on non-transit motorized vehicles at signalized intersections. |
| Definition | The difference between the average time for a vehicle to clear an intersection and the time it would take with free flow speeds. Implemented across the Reporting Period using simplified versions of the definition of delay from Highway Capacity Manual (HCM) chapter 19.3 and using Automated Traffic Signal Performance Metrics (ATSPMs). |
| Necessary Data | ATSPMs calculated by signal management software. |
| Reporting Grain | Intersection, Day Part, and Reporting Period. |
| Iteration Potential | NA |
| Desired Data | NA |
| Considerations | A change in this metric could suggest that TSP is having an impact on non-transit vehicle traffic. Signal performance measurements for non-transit vehicles have been standardized by the Federal Highway Administration (FHWA) and the efforts of various state departments of transportation. However, these metrics may not be calculated in identical manners between jurisdictions and may be unavailable to Caltrans in some jurisdictions. |
Table 7
| Non-Transit Vehicle Segment Travel Time (On-corridor) | Non-Transit Vehicle Segment Travel Time (Cross-Street) | |
|---|---|---|
| Purpose | To measure the impact TSP has on non-transit motorized vehicles along corridors with TSP. | To measure the impact TSP has on non-transit motorized vehicles at signalized intersections intersecting corridors with TSP. |
| Definition | The percentage change in travel time for non-transit vehicles along the project corridor. | The percentage change in travel time for non-transit vehicles on the blocks intersecting the project corridor. |
| Necessary Data | Travel speeds from more precise big data products. | Travel speeds from more precise big data products. |
| Reporting Grain | Day Part and Reporting Period. | Intersection, Day Part and Reporting Period. |
| Refinement Potential | Obtaining access to different data sources for traffic speeds may allow for more accurate calculation of this metric. | Obtaining access to different data sources for traffic speeds may allow for more accurate calculation of this metric. |
| Desired Data for Refinement | More accurate traffic speed data. | More accurate traffic speed data. |
| Considerations | A change in this metric could suggest that TSP is having an impact on non-transit vehicle traffic. However, it is affected by non-TSP factors and may not be reliably calculated for lower-volume corridors. | A change in this metric could suggest that TSP is having an impact on non-transit vehicle traffic. However, it is affected by non-TSP factors and may not be reliably calculated for lower-volume cross streets. |
1. This data warehouse contains archived GTFS real-time and schedule data published by California transit agencies. While Caltrans DDS is developing enhanced data sharing capabilities in this area, other data products based on the warehouse can be found at analysis.dds.dot.ca.gov and public agencies can reach out to hello@calitp.org for requests regarding this data.
| Term | Definition |
|---|---|
| Actual Headway | The actual amount of time between when two transit vehicles arrive at a stop |
| Additional Metrics | A subset of metrics intended to provide context for Primary Metrics and to inform service decisions by transit agencies. |
| Automated Traffic Signal Performance Metrics (ATSPMs) | A set of open-source tools for measuring the effectiveness of signalized intersections, developed through collaborations between various state departments of transportation. |
| Business Rules | Criteria set by organizations to determine operations. In a TSP context, rules set by transit providers and/or jurisdictions to govern behavior of TSP. |
| California Department of Transportation (Caltrans) | The unit within state government responsible for statewide transportation infrastructure and operations, with a mission of “Improving lives and communities through transportation.” |
| Caltrans Data Warehouse | The system used by Caltrans DDS to organize and store processed GTFS Realtime data. |
| Caltrans Division of Data & Digital Services (Caltrans DDS) | A group within Caltrans dedicated to analyzing transportation data and creating data products to better the transportation landscape. |
| Caltrans Speedmaps Tool | A web product that shows transit speed and speed variability for California transit agencies that produce GTFS Realtime. |
| Computer Aided Dispatching / Automated Vehicle Location (CAD/AVL) | A system that provides and keeps track of real-time location data for each vehicle in a transit system. It may also track other vehicle statuses, such as when doors open and close. |
| Day Part | The set of periods over the day for which TSP is reported for. For most Caltrans DDS products, a set of arbitrary but internally consistent "morning peak", "off peak", "evening peak", and "overnight" periods are currently used. |
| Federal Highway Administration (FHWA) | The division of the US Department of Transportation responsible for roadway travel. |
| General Transit Feed Specification (GTFS) | An open standard for transit passenger information data. Comprised of two specifications, one for schedules (GTFS Schedule), and one for real time service information (GTFS Realtime). |
| Granular Trip Time Metrics | Measurements of how a transit vehicle spends time during its trip. These are intended to better diagnose where sources of delay are coming from and support greater levels of optimization. |
| General Transit Feed Specification - Realtime (GTFS Realtime) | A standard for real time transit passenger information data, and a component of the larger General Transit Feed Specification (GTFS). Caltrans DDS has standardized processes to transform GTFS Realtime data into real-time arrivals at transit stops and speeds by segment. |
| Highway Capacity Manual (HCM) | A reference material providing information on various aspects of roadway operations, including conceptual definitions and methodology for performance analysis. |
| Non-Transit Vehicle Metrics | Measurements of the experience of non-transit vehicles. |
| Nth Percentile | The largest number that n percent of elements of a set are less than. For instance, the 20th Percentile travel time of a transit route is the travel time that 20 percent of trips are less than and 80 percent are greater than. |
| Open-Source License | Licenses that allow for software to be freely used, modified, and shared |
| Primary Metrics | A subset of metrics intended to specifically evaluate the performance of TSP systems and support comparison of different implementations. |
| Reporting Period | The duration and frequency at which metrics are reported. |
| Scheduled Headway | The expected amount of time between when two vehicles arrive at a stop based on published schedules. |
| Service Pattern | The exact trip, including route and stops, that a set of transit vehicles take. For instance, if some trips on a route start at one stop, and other trips start at a different stop, they would be considered to form two different Service Patterns. |
| Speed Metrics | Metrics focusing on the absolute speed of transit vehicles. |
| Timepoint | A transit stop which a vehicle is expected to reach at a scheduled time and is not permitted to leave early from. |
| Traffic Signal Controller | A device that causes traffic signals to change their phase based on various inputs, such as signal plans and vehicle detection systems. |
| Transit Cooperative Research Project (TCRP) | A program sponsored by the Federal Transit Administration, known for producing reports helping transit practitioners implement new technologies. |
| Transit Signal Priority (TSP) | A set of technologies that alter the functioning of traffic signals to allow transit vehicles to traverse intersections faster. |
| Trip ID | An ID assigned to a specific daily transit trip. These are used in GTFS to associate trips with information such as its route and the places it stops. In GTFS, they may change over time as schedules are updated. |
| Variability Metrics | Measurements of a transit service’s ability to meet the expectations of its passengers. |
| Vehicle Revenue Hour (VRH) | The number of hours that each vehicle travels while in service. |