Customer case: Enterprise Rent-A-Car
Companies worldwide are struggling to forecast demand due to Covid-19. Enterprise Rent-A-Car started looking for a solution that can help with its supply and demand challenges of their car rental fleet in real-time.
Also see Article in FD Magazine by Peter Ooms.
Enterprise Rent-A-Car is the largest short-term car rental company in Belgium and specializes in vehicle roadside replacement with a fleet of 3,500 vehicles and 40 offices. Enterprise Rent-A-Car has SLAs (service-level agreement) with numerous roadside assistance service, leasing and insurance companies.
“Agreements have become increasingly stringent over the years,” says Christophe Boden, director of Enterprise Rent-A-Car.
Demand is difficult to predict
As with conventional car rental companies, the biggest challenge for Enterprise Rent-A-Car is achieving an optimal occupancy rate.
Christophe explains the challenge “We buy brand new cars and sell them after covering only a few hundred kilometres. In this way, we control the size of the fleet based on expected demand. But our approach differs significantly from that of a traditional car rental company who have a relatively good idea of the need for their fleet, as cars are booked in advance.
Our business is focused on roadside replacement vehicles, which makes us dependent on customers who have either broken down or are involved in a car accident. Subsequently, there is no way of understanding who, which, when and where customers will require a replacement vehicle.”
Until recently, Enterprise Rent-A-Car relied on several statistical calculations based on historical data in its database. “My father’s many years of experience helped us cope with supply and demand. But there was a significant drop due in demand to the pandemic, which had a major effect on our operations. We could not have predicted this. We only succeeded by reducing our fleet, but when the lockdown was lifted, we suddenly needed more cars, and due to the closure of car building facilities, they were short in supply. It also made it very challenging to understand future demand. Needing a solution to help us improve demand planning, we contacted Tangent Works,”
Ten percent more accurate
Christophe continues: “In the fall of 2020, we ran a PoC over three phases with the Tangent Information Modeler (TIM) software and immediately saw the benefits. The predictions were between five and ten per cent more accurate than our system.”
Dirk Michiels, CEO of Tangent Works: “For a more efficient fleet distribution, TIM can also make separate regional forecasts per branch and province. By forecasting demand for each branch, the company can distribute its cars based on TIM’s predictive forecasting capabilities more efficiently.”
The predictions were easily five to ten percent more accurate than what our own system realized. (Christophe Boden, Director)
Christophe continues, “Since the beginning of this year, Enterprise Rent-A-Car has started working with Tangent Works software TIM, Tangent Information Modeler. We included our historical data with external data, such as regional weather forecasts, as diverse weather conditions such as fog, frost, snow, or extreme rainfall historically cause more breakdowns and accidents. Additionally, we looked at the Flemish Traffic Center and other sources for real-time data sources, that could increase TIM’s predictive Machine Learning value. As new data becomes available, TIM can create new forecasting models and forecasts in real time.”
“Another benefit of TIM, is its interpretable AI. We can visualize which features were used by TIM to create forecasting models and predictions and quickly detect and understand new changes in the model. “If the prediction deviates too much from reality, we now see that more quickly and take action accordingly” says Christophe Boden.