The idea that automation always increases efficiency for customers is confounded by the fact that, for some categories of users, it actually makes life harder. In instances where checkout procedures at grocery stores give customers the option of self-checkout or cashier, automation increases the wait time for people who prefer a cashier because automation reduces the number of cashiers by replacing them with self-checkout stations. Coupled with the fact that self-checkout is an imperfect substitute for a cashier, those who prefer cashier can face a longer wait time if there are more people who favor cashier than those who favor self-checkout.

The insight is simple. If self-checkout is a perfect, one-to-one substitute for a cashier, the line at the cashier will stay the same. If more people prefer self-checkout, the line at the cashier can fall. If more people prefer a cashier, then the line at the cashier can increase.

I interviewed 38 people at Walmart and asked them whether they checked out using their automated system, named Scan and Go, or the cashier and why. I visited Walmart during the weekdays between 4 p.m. and 6 p.m., when it was very busy with people getting off work. (Source: https://www.today.com/food/best-time-go-grocery-store-worst-t108655 ). This Walmart had 10 Scan and Go stations and 10 cashiers. I found 53% of people chose cashier and 47% chose Scan and Go. I then broke the data down into those who preferred Scan and Go (42%), those who preferred a cashier (45%), and those who identified Scan and Go and cashier as perfect substitutes (13%).

Those who preferred Scan and Go unanimously said that they chose self-checkout because it was quicker, faster, and more convenient. Those who chose cashier varied more in their reasons for choosing cashier. The reasons included: 1) They had special items including alcohol and prescription drugs; 2) They wanted someone to scan and bag the items for them; 3) They are not computer savvy; 3) They had odd-shaped items; 4) They thought it was faster and easier to use than self-checkout; 5) They had coupons to process; 4) They had a lot of items and the self-checkout station was too small; 5) They wanted to be with someone; 6) “Because I like it,” “Because I want to”; 7) Lazy. Those who said that the line for self-checkout or cashier was shorter as a reason for choosing one over the other were categorized as perfect substitutes.

I interviewed one of the attendants at the self-checkout area and asked him why people preferred Scan and Go. He said that there was always a line at the cashier and there were open, available, stations at the Scan and Go area. So, Scan and Go appears to be faster, but not all the stations were totally utilized because people still prefer cashier.

I turned to the data to see whether self-checkout was really decreasing the amount of time spent at checkout. I looked at the American Time Use survey, which documents how much time people spend grocery shopping during the week versus during the weekend and holidays when the lines at the grocery store are the longest. (Incidentally, lines during weekdays between 4 p.m. and 6 p.m. are also long because of people shopping when they get off of work.) (Source: https://www.today.com/food/best-time-go-grocery-store-worst-t108655).

I took the time spent on the weekends grocery shopping minus the time spent on weekdays and looked to see if the difference decreased or increased over time. The idea is that the longer wait times on the weekends would drive people to use self-checkout more than usual. In other words, people on the threshold between self-checkout and cashier would be nudged toward self-checkout when the lines on the weekends are longer. Also, looking at the difference in shopping times during the weekdays and weekends for the same person automatically controls for idiosyncrasies between people.

I found two categories of respondents for whom self-checkout seems to have decreased the time spent at the grocery store–those age 55 to 64 and those age 15 to 24. The plot of the difference in time spent on weekends and time spent on weekdays grocery shopping for the 55 to 64 demographic looks like this:

The trend line is slightly downward sloping and a quick calculation shows that the difference in time spent grocery shopping between the weekend and weekdays fell by 0.33 minutes, or 19.85 seconds, between 2003 and 2018.

The graph for the 15 to 24 age group is even steeper and looks like this:

The trend line is downward-sloping and shows that, between 2003 and 2018, the difference in time spent grocery shopping between the weekends and weekdays fell by 1.39 minutes. This corroborates the story that younger age groups are quicker to adapt new technologies than older cohorts.

The age categories in between 24 and 55 all faced increased wait times at checkout. For example, the graph for the age group 45 to 54 looks like this:

Between 2003 and 2013, the trend line shows that the difference in checkout time between weekend and weekdays increased by 1.99 minutes. The other age cohorts, 35 to 44 and 25 to 34 all look the same. For those age 35 to 44, the difference in shopping times increased by 1.31 minutes. For those age 25 to 34, the difference in shopping times increased by 0.61 minutes.

Why did wait times for these cohorts increase? It has to do with having children under 18. My interpretation of the data is that people with children prefer using the cashier over self-checkout because it is more inconvenient to scan and bag groceries with a child in tow.

The data bear this out. The following chart shows differences in grocery time between the weekend and the weekday for people with and without children under 18:

The trend line for “Difference Children,” which captures people with children under 18, is increasing and shows that, between 2003 and 2018, the difference in grocery shopping times between the weekend and the weekday increased by 2 minutes. For “Difference No Children,” which captures people without children under 18, the trend line is basically flat–an insignificant increase of 5.56 seconds. Therefore, for the cohorts 25-34, 35-44, and 45-54, cohorts who are most likely to have children under 18, the wait time at the cashier has increased the amount of time spent grocery shopping.

The idea that automation naturally increases the speed of checkout times is confounded by having an imperfect substitute between automated and legacy systems of checkout, and having a partial deployment of automated systems, so that people have a choice between the two. This article demonstrates that some age cohorts have seen the time spent grocery shopping decrease by as much as 1.31 minutes, while other age cohorts with children have seen time spent grocery shopping increase by as much as 1.99 minutes. The impact of automation on checkout times is not unanimously time-saving.

Source: American Time Use Survey ( https://beta.bls.gov/dataQuery/find?st=0&r=100&s=popularity%3AD&q=grocery&fq=survey:[tu]&more=0)