Data Science opportunities for ground transportation companies to improve Last Mile Delivery operation in the era of Ecommerce.

Last Mile Delivery in the era of Ecommerce

In a products journey from the manufacturing company warehouse to the customer’s hands the last mile delivery refers to the last part of the shipping process (Vakulenko et al., 2019). Before the big surge of Ecommerce, the last mile delivery usually took place at a business, and deliveries consisted of consolidated orders moving on pallets or crates. People would later go to these businesses to buy products; they did not have contact with the ground transportation company and therefore their shopping experience was not affected by the shipping process (Jiang & Rosenbloom, 2005). In contrast, after the surge of the Ecommerce era, the Last Mile delivery often takes place at residential areas and consists of individual orders in the form of packages, the final consumer of the products has direct contact with the ground transportation company. The experience they have with the whole shipping process will highly determine the level of satisfaction they got from the complete shopping experience.

The growth of e-commerce has influenced on a grand scale the development of retail and logistics industries in recent years (Vakulenko et al., 2019). Statista (2021) reports that during 2020, the retail Ecommerce sector in the United States reported sales for US$792 billion, up by almost one third compared to 2019, by 2025 it is expected to reach US$1.65 trillion. While retail in-store shopping is still going strong, it is undeniable that the retail sector is going through a shift to online shopping that has brought multiple challenges and opportunities for companies in the ground transportation industry that are facing customers with high expectations and a wide array of logistic options.

The Ground transportation industry collects a high amount of data throughout the entire shipping process, from routes travelled, to location facilities where the package spent time, the trucks transported the packages, and the millions of consumers who buy products online and ship them to their homes. Additionally, the technology behind the Transportation Management Systems from every major actor in the supply chain, the big advances in technology such as real-time GPS tracking, and the high availability of datasets, APIs and open-source platforms provide high value and multiple opportunities for Ground transportation companies to innovate through Data Science.

This paper discusses different challenges ground transportation companies are facing to serve retail Ecommerce customers, and how innovation through data science practices can help them tackle those challenges, while generating a positive impact. The last mile delivery has proven to be complicated, time-consuming, and expensive, for that reason we are going to discuss solely this part of the shipping process and will focus on the retail Ecommerce sector given its exponential growth tendency.

Challenge: Reducing costs and improving the overall customer experience

The last mile is known to be the most cost-intensive part of the supply chain (Vakulenko et al., 2019), Yrjölä (2001) also shows that the last-mile delivery cost can represent about 50% of the total logistic costs, it is highly sensible to the factors of the delivery and can be reduced though the implementation of proper strategies. So, what makes the last mile delivery so complicated and expensive?

While most of the ground transportation process takes place in highways and deliveries in distribution centres, the last mile delivery takes place in urban and rural areas, most predominantly in the former (Ozarik et al., 2021), both presenting challenges and implicit costs. Rural delivery stops might be far apart from each other, with few packages being delivered at each location, increasing the number of miles travelled per delivery and fuel consumption. In urban areas even though the delivery stops are usually closer to each other, additional factors such as traffic congestions, roads closures, traffic accidents, represent a challenge and add up to the cost in the form of fuel efficiency, time, among other factors (Janjevic et al., 2019). Costs resulting from inefficiencies continue to rise as demand grows. Customers expect fast, transparent, and cost-effective shipping processes, but they also expect a high-quality service and an overall good experience.

In order to successfully address the expectations of online customers, recent e-commerce and retail research have developed a comprehensive customer experience agenda (Pandey and Chawla 2018). Research shows an increase in customer demands for service quality, customers expect to receive a personalized service with flexible delivery options, and more convenient methods of collecting and making returns (Michałowska, Kotylak, & Danielak, 2015; UPS, 2015). Customers also expect full visibility and transparency during the shipping process. Additionally, they want rapid access to information, communication channels with the ground transportation company and easy processes in the case of eventualities. There is evidence that the level of customer satisfaction can vary significantly from the stages of buying and after delivery, suggesting that delivery has a great impact on overall satisfaction (Jiang & Rosenbloom, 2005).

So how can data science help companies in the ground transportation industry reduce costs in the Last Mile Delivery process while improving their customer experience for the demanding customer of the Ecommerce era?

Optimize route planning and driver assignments

Route planning is a fundamental problem in last-mile delivery, which consist of scheduling multiple couriers’ routes sequences of origins and destinations locations under determined optimization objectives (Zeng et al., 2019). A good way of reducing costs is improving efficiency. From the moment a pick-up order is received, ground transportation companies need to assign that order to a driver, who will also deliver other orders during the day according to the route provided by the company. This is where the first data science opportunity to reduce costs come.

By creating real-time route optimization with location intelligence, ground transportation companies can improve their performance by using Spatial Data Science techniques that allow them to build data models to simulate existing conditions, giving information about constraints, inefficiencies, among others (Zeng et al., 2019). To do so, first they need to create an origin-destination cost matrix and then create an algorithm to process the orders from the system and assign them to a driver based on their location and any other additional factor such as pending deliveries, expected traffic, weather conditions, etc. To feed the algorithm, the company will use data from historic shipments, forecasts, Geolocation of drivers, distribution centres and delivery points, traffic congestions, road closures, among others. By Automating route optimization and driver assignments, ground transportation companies can save time and perform more deliveries with the same number of trucks and drivers, with less driving time and less constraints and less costs.

Another important challenge when defining routes and driver assignments is unsuccessful deliveries and returns. A major problem in last-mile delivery is the high rate of unsuccessful deliveries caused by the probabilistic absence of customers (Ozarik et al., 2021). From the Data Science perspective, the goal is to reduce costs resulting from unsuccessful deliveries by considering routing and scheduling decisions simultaneously. Computational experiments show that using customer-related presence data significantly can yield savings as large as 40% when compared to traditional vehicle routing solutions (Ozarik et al., 2021).

Some of the issues in implementing this kind of innovation is that it can become hard to improve them when the model requires complex features, also the amount of data from different sources can make it difficult to work when there is missing or failing data.

Fleet Management through IoT

The Internet of Things (IoT) and big data have been hot topics in recent years. With the amount of data being produced, new applications like predictive maintenance which attempts to predict the health of equipment using machine learning are now a reality (Kileen et al., 2019). For ground transportation companies with owned fleet of vehicles, the Internet of Things provide a great opportunity to improve their vehicle’s performance, maintenance, driver operation, and cargo management. Implementing telematic solutions on the vehicles in the form of sensors can help companies identify the best moments to perform oil change, tire replacement, change of breaks, battery replacement, etc. These sensors collect data about the vehicle use and performance and help companies save costs by taking a preventive approach to their fleet management, reducing costs by preventing vehicle breakdowns and major loss of time, money, and deliveries. By applying machine learning models to the data collected by the sensors and the historic data from previous fleet management operations, companies can identify factors that would require a vehicle to receive maintenance in advance, schedule maintenances in a way that does not affect the capacity to attend orders by having multiple vehicles out of operation, it lowers costs by anticipating failures, unscheduled maintenance, and downtime, and by making sure that failing parts are replace only when actually needed (Furch et al., 2017).

Crowdsourcing for same day deliveries

The advances of the sharing economy and communications technology has made the crowdsourcing concept to become more popular in recent years. Uber and Airbnb are good examples. Crowdsourced transportation has an important role when facing the growing demand for last-mile delivery, most predominantly since the surge of e-commerce, one of its benefits is that provides more flexibility and demands less capital investment than traditional outsourcing approaches. (Huang & Ardiansyah, 2019). Even though in Logistics the practice of outsourcing services has been widely used, most notably in the form of freight brokerage, the concept of crowd-delivery only became popular recently (Rouges & Montreuil, 2014).

Among the benefits of using crowdsource transportation we can find an increase in fleet utilization, a decrease in vehicle mileage, and a cost saving especially for last-mile delivery (Huang & Ardiansyah, 2019). As ground transportation companies plan their routes in advance, same day deliveries are particularly complicated as it might be difficult to find an available truck or driver to perform the delivery. This is where the crowdsourcing model can provide additional value.

Similar to the optimization of route planning discussed before, the crowdsourcing model would use a variety of spatial data science techniques to assign drivers for same day deliveries, this drivers will take orders based on their availability, proximity to the pick up and delivery location, performance score obtained by the driver in previous operations (Zeng et al., 2019). By mixing their own fleet, with outsourced vehicles and crowdsourced drivers, ground transportation companies can provide more efficient solutions to their customers based on their need and the time sensitivity of their order.

The main issue in implementing crowdsourcing deliveries lies on the fact of working with external and sometimes informal drivers, most predominantly in the case of inconveniences such as package or product damage. The customer experience can be negatively affected by an external actor.

Conclusion

As complicated and cost intensive as the last mile delivery has proven to be, is it undeniable that there are big opportunities for improving costs and efficiency which will translate into a better customer experience. Companies who are prepared to take advantage of major opportunities in the field of data science as the ones mentioned in this document can generate a competitive advantage that will allow them to better serve their customer while getting better economic results.

References

Yulia Vakulenko, Poja Shams, Daniel Hellström & Klas Hjort. 2019. Online retail experience and customer satisfaction: the mediating role of last mile delivery, The International Review of Retail, Distribution and Consumer Research, 29:3, 306–320, DOI: 10.1080/09593969.2019.1598466. https://doi.org/10.1080/09593969.2019.1598466

Milena Janjevic, Matthias Winkenbach, Daniel Merchán. 2019. Integrating collection-and-delivery points in the strategic design of urban last-mile e-commerce distribution networks. Transportation Research Part E: Logistics and Transportation Review. https://doi.org/10.1016/j.tre.2019.09.001

Shelagh Dolan. 2021. The challenges of last mile delivery logistics and the tech solutions cutting costs in the final mile. https://www.businessinsider.com/last-mile-delivery-shipping-explained?r=AU&IR=Thttps://www.businessinsider.com/last-mile-delivery-shipping-explained?r=AU&IR=T

Yuxiang Zeng, Yongxing Tong, Lei Chen. 2019. Last-mile delivery made practical: an efficient route planning framework with theoretical guarantees. Proceedings of the VLDB EndowmentVolume 13Issue 3November 2019 pp 320–333https://doi-org.ezproxy.lib.uts.edu.au/10.14778/3368289.3368297

Sami Serkan Ozarık, Lucas P. Veelenturf, Tom Van Woensel, Gilbert Laporte. 2021. Optimizing e-commerce last-mile vehicle routing and scheduling under uncertain customer presence. https://doi.org/10.1016/j.tre.2021.102263

H. Yrjölä. 2001. Physical distribution considerations for electronic grocery shopping. Int. J. Phys. Distrib. Logist.

Miguel Alvarez Garcia. 2019. On-demand last mile transportation. https://towardsdatascience.com/on-demand-last-mile-transportation-ea4a3b725b31

Kuancheng Huang, Muhammad Nashir Ardiansyah. 2019. A decision model for last-mile delivery planning with crowdsourcing integration. https://doi.org/10.1016/j.cie.2019.06.059

J.-F. Rouges, B. Montreuil. 2014. Crowdsourcing delivery: New interconnected business models to reinvent delivery 1st International physical internet conference.

Furch, J., Turo, T., Krobot, Z., Stastny, J. 2017. Using telemetry for maintenance of special military vehicles, in: International Conference on Modelling and Simulation for Autonomous Systems, Springer.

Patrick Killeena, Bo Dingb, Iluju Kiringaa, Tet Yeapa. 2019. IoT-based predictive maintenance for fleet management. The 2nd International Conference on Emerging Data and Industry 4.0 (EDI40). https://doi.org/10.1016/j.procs.2019.04.184