top of page

Walmart using employees for last mile delivery? Baaaad idea.

One more try at the last mile

Last year, Walmart launched a radical plan to fight online delivery services. The idea was to use their own store employees to bring, after completing their shifts, online orders to customers’ homes. Walmart was expecting to tap into its massive workforce to lower the “last mile” delivery costs.

Months later, Walmart has had to quietly shut down its initial pilot program, according to Reuters.

This is just one of the multiple attempts from Walmart to get the “last mile” right. Other plans have been motorbike delivery in Mexico, using small supermarkets in China as micro-depots to deliver goods within 30 minutes, and the conventional warehouse to customer design in Japan, though this last attempt in Japan seems to be yet another failure of a foreign retailer failing in this country.

What went wrong?

Speaking with a good friend and advisor Danielle Beauparlant Moser, we both agreed that, for an organization of this size, the pilot seemed ill-conceived and unsophisticated given they were 'going after' Amazon. What went wrong?

Retail, Last Mile Logistics
Credit: ElasticComputeFarm


One issue was compensation: Walmart was paying employees (associates in Walmart’s’ parlance) for their time and costs, on top of their shift hourly pay, which starts at $11 an hour. Reuters reports that associates where paid $2 a package, plus 54 cents for fuel per mile, and an extra hour of overtime pay. Fair? The driver was going on their way home, so they were already incurring in part of the millage cost. But…

No, this is not Uber

The problem is thinking that goods delivery, and particularly grocery delivery, is like “Uber”. Uber’s operational design is complex, but not because of its routing: a relatively simple shortest path from A to B (or A to B to C in some forms of ride sharing options), of a self-propelled, non-perishable, intelligent, interchangeable item: a person.

The first problem was dispatching: employees reported having to wait up to 30 minutes for the store to have the items ready after the end of their shift. This time already was consuming half of their extra hour remuneration. Optimal dispatch, particularly in extremely dynamic environments like this one, with orders placed continuously over time through Walmart’s online channel, and routes that have to be adapted to the home addresses of their enrolled employees, is not easy. Additionally, dispatch needs to consider weight and volume limitation, particularly difficult in this environment, using non-standardized, non-commercial vehicles.

Combined with the dispatch is the routing problem. Orders have to be linked with the departure point (store) and final route point (employee’s home). Customers’ time windows (period of time allowed by the customer to receive the package), and a narrowly defined startups time (employees’ end of shift) further constrains route optimization.

Unlike Uber, were the customer is dropped off in the sidewalk incurring in an almost negligible stop time, goods need to be delivered home, particularly in this case, where there are perishable items. There needs to be a proof of delivery. Despite being a straightforward task, part of the advantage of giants like USPS or Fedex comes from their drivers’ knowledge (and sometimes control) of the last meters of the last mile. I would not be surprised if stops were exceeding 10 minutes of service time for Walmart’s untrained staff. If the employees were typically doing about 5 packages a day, that is already 50 minutes on top of the 30 minutes dispatch. If the drivers had to add 10 miles more to their daily commute to service these 5 clients, add another 20-30 minutes of urban/extra-urban driving time. Rounding up, this is an additional 2-hour task, paid at roughly $23 ($13 extra hour plus $2 per package, 5 packages), considering that the millage covers the vehicles additional costs for the extra 10 miles. This is $11.5 an hour, below their regular hourly pay. “The money was never worth it… and they stopped paying mileage in the end” an anonymous employee told Reuters.

Add the hidden costs

Worse of all, there are hidden costs that probably did not emerge during the short pilot test. According to the reports, the program was unclear about insurance and risk transfer between store, employees and customers. I am convinced that customers considered their home and their signature on the proof of delivery the place and time of the transfer. What happens during the transport, in an unprepared, unrefrigerated vehicle insured for only for private use?

The underlying inefficiency

The problem is, in my view, purely and simply poor operational design. Through the use of the return miles of employees driving home may sound attractive to save costs, the restrictions in logistic design are too great. A well-operated last mile drop-off urban operation can easily exceed 30 stops considering a 10-minute service time in an 8-hour shift, including dispatch and return to depot. This is more than 3.8 orders an hour compared with the 2.5 orders per hour of an employee driving home. In fleet operations, optimizing jointly routes of multiple vehicles, orders are so well dispatched and routed that most of the driver’s time is consumed in the stops, not moving between stops. 

What is the solution, then?

There is no doubt in my mind that optimization of last-mile logistics require dedicated fleets and tight central planning and monitoring, so I am not very optimistic about freelancer-uber-type goods delivery designs.

In any case, tech applied to the decision-making processes involved in the delivery of goods will become increasingly important, particularly if companies like Walmart want to make use of legacy assets, like using urban shops as micro-depots. At Innitium we have seen how advanced algorithms can achieve easily double digit increases in efficiency. But higher supply chain complexity implies more sophisticated decision-making support systems and wiser strategies, as Walmart has learnt the hard way.

34 views0 comments


bottom of page