Public Transportation

Developing a Predictive Model for Service Delivery



Lockdown and social distancing measures to contain the spread of COVID-19 created a severe economic downturn world over. In the midst of this uncertainty, businesses needed to grapple with a myriad of questions: What will the re-opening of the economy look like? Who will be our first customers? How will we meet the needs of our customers given a decrease in demand and a fear of proximity? To help answer these questions we built a predictive model of customer behavior with easy to follow steps to test assumptions.

The Challenge

The challenge was to plan for a post-pandemic world, given the uncertainty and the possibility of many future scenarios.

The Process

To tackle this challenge, the team began by defining customer profiles using historical data. With stakeholders from across the business, a scenario planning exercise was conducted to define potential business outcomes. The key indicators and the potential impacts of the scenarios were defined that were used in the predictive models. In building the model, some assumptions had to be made about human behaviour. Keeping in mind that these assumptions may not hold true in an economy reawakening from lockdown, we identified steps the client can follow to test these assumptions and dive deeper to gain more insight. The combination of qualitative activities and quantitative modelling will help the client adjust to a rapidly changing environment and modify their approach if required. This upfront iterative planning will equip our client with the necessary tools needed to navigate the uncertainty.

The Output

The output was a predictive model for service delivery and prioritization for essential workers in a post-pandemic world, that was backed by research and based on possible scenarios. Qualitative activities were recommended to test the model assumptions and provide flexibility in the event that plans needed modification.

Keys to Success

  • Scenario planning
  • Customer Segmentation
  • Predictive Behaviour Modelling