The Journal of Business Logistics (JBL), published by the Council of Supply Chain Management Professionals (CSCMP), is recognized as one of the world's leading academic supply chain journals. But sometimes it may be hard for practitioners to see how the research presented in its pages applies to what they do on a day-to-day basis. To help bridge that gap, CSCMP's Supply Chain Quarterly challenges the authors of selected JBL articles to explain the real-world applications of their academic work.
"Crowdsourcing Last Mile Delivery: Strategic Implications and Future Research Directions," by Vincent E. Castillo of The Ohio State University, John E. Bell of the University of Tennessee, William J. Rose of University College Dublin, and Alexandre M. Rodrigues of the University of Tennessee. Published in the December 2017 issue of the Journal of Business Logistics.
As e-commerce sales continue to grow, last-mile delivery has become increasingly important to retailers. In response, many companies have started experimenting with "crowdsourced logistics" (CSL) to fulfill their customers' desire for fast, on-demand delivery. This sharing economy model patterns itself after ride-sharing services such as Uber or Lyft, but instead of transporting people, the drivers transport goods.
Because the sharing economy model is so new, it's not yet clear just how effective CSL is as a delivery strategy. For example, when companies work with a driver on an individual contract basis, they expose themselves to more risk and uncertainty than when they have a dedicated fleet of full-time drivers. That's because, under the sharing economy model, drivers manage their own schedules and work as long or as little as they desire. Therefore, companies cannot be certain of the supply of drivers that will be available to them at a particular time.
To get a better idea of how CSL compares to a dedicated fleet, the article's authors designed a simulation model of delivery services from an Amazon distribution center to 1,000 customer locations throughout New York City. The model compared the logistics effectiveness of a traditional dedicated fleet of delivery drivers to the use of crowdsourced logistics.
The article's lead author, Vince Castillo explained to Supply Chain Quarterly Executive Editor Susan K. Lacefield what the model revealed about crowdsourced logistics and how companies can apply these findings.
Why were you interested in studying crowdsourced logistics?
We wanted to study the last-mile delivery version of crowdsourced logistics for a few reasons. First, at that point in time, most of the academic literature was focused on the ridesharing model that moves people rather than goods, so there was an opportunity to try to build knowledge about and draw attention to this phenomenon that was emerging in practice. Second, the topic is one that we thought both practitioners and academics would be interested in. This meant that as long as we could develop a rigorous study, it would definitely have relevance, and both of those things are required to make worthwhile contributions. Finally, with the continued growth of e-commerce, the importance of last-mile delivery, and the impression that crowdsourced last-mile delivery could be a scalable solution, we felt this was a timely study to undertake.
Why did you decide to use a simulation model to look at the logistics effectiveness of crowdsourced logistics vs. a more traditional fleet of delivery drivers and vehicles?
Being such a novel innovation for delivery, we wanted to learn if and how CSL affects a shipper's last-mile strategy and more generally, its supply chain strategy. To answer these questions and to understand how and when CSL could be used in practice, we felt it was important to first understand the capabilities of a crowdsourced fleet in terms of logistics effectiveness. But for those capabilities to make any sense, we needed a comparative baseline, which is why we chose to think about CSL's effectiveness relative to a fleet of dedicated delivery agents. We were hoping to find differences in logistics effectiveness between the two fleet types that we could use to build middle-range theory about the contexts in which CSL might be used.
What results did the model show, and were any of them surprising to you?
We had some results that were somewhat counterintuitive and rather surprising. There are a number of differences between crowdsourced and dedicated fleet types that shippers have to consider when using CSL. We focused on one new variable in this study—the uncertainty in a supply of crowdsourced drivers that emanates from their autonomy. It's this autonomy of gig economy workers that intrigued us because it is common to all types of services that can be crowdsourced, so our findings could feasibly be more generalized. By looking first at the uncertainty in the availability of a supply of crowdsourced drivers, we expected that effectiveness in terms of total deliveries and on-time delivery rate would be lower for CSL than in a dedicated fleet of drivers across a number of delivery scenarios. Our hypotheses in these cases were mostly supported, and we confirmed that most of the time, dedicated is likely to be more effective than crowdsourced delivery.
However, when we increased delivery demand intensity (increasing the number of orders received and with less time between order receipts), we found that there were cases in which CSL was actually more effective than the dedicated fleet in terms of making more total deliveries. This was one of the surprising results because we expected that a dedicated fleet with known capacity and availability would always outperform the crowdsourced fleet comprised of amateurs who may or may not accept deliveries they're offered. It turns out though that when the fixed-size dedicated fleet reaches maximum utilization, additional delivery requests received beyond the capacity of that dedicated fleet are more likely to be late or even rejected, potentially meaning lost sales. Thus, fewer deliveries can be made because of the fixed capacity if the dedicated fleet is too busy. CSL, on the other hand, doesn't have the same upper boundary on its capacity, so a company could activate a crowdsourced fleet in the event demand starts rising above a certain level to respond in kind and perhaps not lose out on any sales.
How could practitioners apply your research?
I would say that practitioners interested in crowdsourcing last-mile delivery should recognize that this research highlights some of the nuance that they need to understand before employing this business model. CSL is not a panacea for last-mile delivery, and I don't recommend that anyone doing home delivery go and cancel their dedicated delivery contracts in favor of a fully crowdsourced last-mile strategy. However, there are other benefits, namely in the use of CSL as a backup plan to be able to serve delivery demand when it surges unexpectedly. That is, CSL appears to be a way of increasing agility and responsiveness in the last mile of the supply chain. Furthermore, CSL could also be used where delivery time windows are not critical to customer service—like in the case of online returns. These two applications need more research though, which we are currently undertaking.
For this particular study you looked at an Amazon distribution center in Manhattan. Do you think your results would have been significantly different if you had used a different kind of company or a different location?
Yes, and in fact, if you look at other types of companies that are crowdsourcing last-mile delivery, the product type seems to make a difference. For instance, shipping groceries and meals from local restaurants have been some of the more successful ventures, while a startup that crowdsourced flower delivery recently went out of business. Now there could be any number of reasons that the latter firm did not succeed, but the product seems to be important, namely because different products have different demand predictability, which affects intensity of on-demand delivery orders received.
I would also expect that in practice, CSL's effectiveness would differ across cities. For CSL to be somewhat reliable, you have to be near a population that is amenable to working in the gig economy. That mostly exists in large cities for the time being, although there are an increasing number of citizens from rural areas interested in working gig economy jobs. Furthermore, cities are designed differently with some being more conducive to logistics traffic than others as well as having different policies and regulations to account for. It's a question certainly worth exploring more deeply.
What do you see as the key takeaway message?
At first glance, it may seem like the draw of CSL is that it is cheaper than dedicated delivery, but this isn't necessarily the case. Companies typically guarantee an hourly wage or pay drivers by the mile on top of a per delivery wage. The primary benefit of CSL is actually increased agility, responsiveness, and flexibility in the last mile, which goes a long way to increasing repeat- purchase behavior and customer-service quality. Furthermore, crowdsourcing provides the potential for shippers to acquire a scalable last-mile delivery solution that has a much higher capacity than a dedicated fleet...if they can find the right formula that works for them.