National Highways Incident Delay Metric


City Science was appointed by the Network Analysis and Statistics team of National Highways to develop a new metric to measure delay on the Strategic Road Network caused by Incidents, accompanying software for calculating the metric’s value, and a predictive model for forecasting its value. The new metric is a potential option for reporting by for National Highways to its customers and stakeholders during the third road investment strategy period (RIS3), which is due to run from 2025 to 2030.



National Highways committed to develop a new metric to measure delay associated with incidents arising on the entire Strategic Road Network during the Road Investment Strategy 2 (RIS2), which runs from 2020 to 2025. The purpose of developing this new metric is to align with the long-term ambition of National Highways, the DfT, Transport Focus and Office of Rail and Road (ORR) to measure the outcomes of incident management on customer journeys.

National Highways commissioned City Science to progress development of the metric. The key elements of our scope included:

  • Identification, appraisal and recommendation of data sources to use within the metric including incident detection and capturing the full extent of delays associated with incidents.
  • Selection of a metric and communication method of explaining the metric reporting option.
  • Creation of a process and algorithm for calculating the metric’s value.
  • Development of a forecast model to predict future metric performance.

City Science Response

Stage 1 – Metric Review: We initially reviewed previous work related to the incident delay metric during the concept and feasibility stages. This was complemented by a review of other similar delay-related metrics either established or in development. In parallel, we engaged extensively across departments within National Highways (e.g. operations, customer service delivery) alongside Transport Focus, the Department of Transport and ORR to establish key metric success criteria. The outcomes of this process informed research recommendations for consideration in Stage 2.

Stage 2 – Metric Research & Development: Stage 2 of the process involved:

  • Definition Development: We defined key terms such as ‘Incident’ and ‘Delay’ to inform the metric development process, which was developed in partnership with key stakeholders. Accounting for feedback we received from stakeholders, our definition of an ‘incident’ focused on unplanned events.
  • Data Source Appraisal & Selection: We completed a detailed review and appraisal of possible data sources to use within the metric both to capture incidents and associated delay. We considered a wide range of sources such as INRIX, TomTom, ControlWorks and Waze and appraised the sources’ relative robustness, futureproofing, cost-effectiveness and representation across the Strategic Road Network.
  • Data Engineering & Algorithm Development: We developed a robust data analysis algorithm for detecting and verifying potential incidents and capturing associated delay using our selected data sources (TomTom and ControlWorks). Our algorithm was developed to work across every road link at one-minute intervals across the entire 4,500-mile Strategic Road Network. We worked closely with National Highways to ensure our algorithm could work directly on National Highways internal data systems.
  • Metric Reporting & Communication: Through hosting a series of workshops with key stakeholders, we developed, appraised and selected a preferred metric reporting option. We developed a clear communication method for explaining and reporting the metric to customers and wider stakeholders. This was aided by visual graphics and customer journey example anecdotes.
Our Metric Development Process
Our Metric Development Process

Stage 3 – Predictive Model Development: We developed a predictive model using an entire year of historical 2019 data so National Highways could understand the potential future performance and relative sensitivity of the metric. We developed and validated our model which was then able to assess the impact of various input variables including changes in traffic flows and seasonality on incident delay.

Model Outputs
Model Outputs Showing Number of Incidents Based on Traffic Flow
Model Validation Graphs
Model Validation Outputs Comparing Predicted Delay Against Actual Delay (Left) and Predicted Incidents with Recorded Incidents (Right)


This was a challenging project due to the large volume of data involved, the difficulties in reconciling data streams, the complexities in identifying incidents and attributing delay, and building consensus around how delay should be communicated.

However, we successfully overcame these challenges and produced a high-quality report for National Highways which consolidated our findings. We presented these to stakeholders such as the DfT, ORR and Transport Focus alongside National Highways’ internal Metric Assurance Group, who are responsible for quality assuring in-development metrics. The Metric Assurance Group recommended that our metric be progressed by National Highways to the next stage of development: the validation stage.