The Psychology and Economics of Line-Balks

You thought I was done with my sequence of articles on Starbucks … you were wrong.

I thought I was done with ranting about Starbucks … I was wrong as well. 

Because when Brady Brewer, Chief Marketing Officer of Starbucks, made the following statement at the company’s investor day on September 13th in Seattle, I just had to write about it:

“Long drive-thru lines are ‘one of the number one reasons for not visiting Starbucks,’ Potential customers see the long line from the road and decide not to join, or leave from the end of the line before ordering, which is referred to as ‘line balk,’ at Starbucks.”

To paraphrase Snoop Dog: “You can take the boy out of Starbucks, but you cannot take Starbucks out of the boy,” or something like that. 

But this is not only a Starbucks thing. In 2021, Dan Cathy, the CEO of Chick-fil-A at that time, stated the following: 

“We estimate about 30% of the people are driving off, driving away, because the lines are so long,”

These statements are also somewhat related to the maxim that exists in the Quick Service Restaurants industry, that 7 seconds improvement in waiting time results in about 1% of market share. 

When you ask the people who make these claims how data-driven they are, you usually get a very hand-wavy response. It should be clear that estimated inaction in the physical world (unlike online) is quite hard. We don’t really know who decides not to join the line. We can impute it, but it’s not a trivial exercise. 

In fact, there are multiple questions that these statements raise:

1. Do people care about time? We’re all pretty sure they do. But do they? 

2. If so, can we measure the extent to which they care? 

3. And finally, what’s more important: the length of the line or the actual anticipated waiting time? 

The Impact of Waiting Times

So yes. Customers care about speed. In my own work on the topic, focusing on drive-thrus, we show that people behave as if their cost of waiting is approximately ten times the average wage and about 30 times the minimum wage. In other words, our study showed that people tend to hate waiting so much that they believe their time is much more expensive than it really is. So, bottom line, people really really hate waiting.

On a related note, in the same study, travel time to the location didn’t demonstrate a similar extreme reaction. In other words, people were much more rational in choosing a location based on proximity and almost didn’t treat the travel time as time wasted. 

So yes, people care about time. And yes, it’s possible to estimate how much they respond to waiting times (even with limited data), and it was a major estimation endeavor that took us almost four years of work. So when I hear statements on the impact of waiting times, I’m usually skeptical. The impact of 7 seconds improvement was in fact, between 0.5% to 2.5%, and highly dependent on the chain. 

Many years later, I was an expert witness in court in Alabama, for a dispute between a burger chain and a soft drinks brand. I’m not sure I can say more, other than the fact that I watched the movie My Cousin Vinny in preparation for my deposition. 

The Impact of Queue Length

But note that both Starbucks and Chick-fil-A mention line balks and not the impact of waiting-time per se. Can we estimate the impact and extent of those? And what’s more important? The length of the line or the waiting time? 

A research paper (conducted at the deli stand of a supermarket) by Marcelo Olivares, Andrés Musalem, and their co-authors, yields interesting results, showing that:

“...purchase incidence is mainly affected by the number of customers in line, and does not seem to be affected by the level of staffing of the queue. This is consistent with customers using the number of people waiting in line as the only visible cue to assess the expected waiting time, not fully accounting for the actual speed at which the line moves.”

One of the most interesting graphs from the paper is:

The graph shows that, in fact, when the length of the line increases, people are more likely to join. Why? Because of herding. People like seeing that other people like the place. No one wants to dine in empty restaurants. No one wants to join a nonexistent line. 

But beyond a certain point, the impact of the purchase incident decreases with the length of the line and can go from around 32% to 22%, which is around 30% —very much what the CEO of Chick-fil-A claimed!

So people indeed react to the anticipated waiting time (my study), and when they arrive and want to decide whether they will join or not, they respond to the length of the line more than the speed at which it progresses. 

And clearly, this is more acute when there are drive-thrus, since the line is more salient: it’s clear how many people are in front of you in the line —more than in the “cloud-based” line that is typical at Starbucks. 

Starbucks claims that about 50% of sales are currently from drive-thrus, and as I mentioned before, 90% of their new locations will have drive-thru service, including drive-thru only locations, given their new emphasis on speed. 

The Solutions?

One of the main initiatives Starbucks is taking is the use of handheld ordering devices to take orders while people stand in line. 

They claim that these “handheld ordering devices break up the ‘huge bottlenecks’ of traditional drive-thrus.” But is that the case?

“Chick-fil-A has similar methods for making drive-thrus faster and more efficient, like installing double drive-thru lanes and having workers take customers’ orders on tablets at their cars before they reach windows.” 

If you’re wondering what dual lines are:

“The dual lines will help stack the vehicles in the Chick-fil-A parking lot, instead of having the line push out into the road through the shopping center. The two lines will have separate ordering systems, but the lanes will merge prior to reaching the window to receive your food.”

The key question is whether these initiatives actually help. 

It’s easy to see that they help if:

1. The order taking is indeed the bottleneck (and not the food prep);

2. People respond to the length of the queue rather than the total number of people waiting;

3. Breaking down the waiting time to wait-time-until-you-order and wait-time-until-you-get-the-order helps. 

These initiatives don’t help if:

1. The kitchen (and food prep) is the bottleneck, since the overall sojourn time will then be the same; 

2. People account for the total volume of cars rather than any specific line; 

3. People don’t mind waiting, but they don’t want the waiting-time to be broken down. 

So which one is it?

For Starbucks, it’s clear that making the orders is the bottleneck, which is also probably the case for Chick-fil-A. 

Whether people respond to the number of cars waiting or the length of the line is an empirical question I don’t have an answer to, but I can see why a long line would be more deterring than several shorter lines, even if the total waiting time is similar. We know from previous research that people are “happier” standing in a line that moves fast, and the reality is that a single long line moves faster than two shorter lines (given that both funnel to the same pick up point). But this level of satisfaction assumes that people join the line. If people only respond to the length, a long line is going to be more deterring than two shorter ones. 

But this brings me to the final question: Is there value in breaking down the service, and if so, is doing this early (i.e., having workers wander around with tablets) helpful? 

Here, Carmon and his co-authors have an interesting answer. They show that many times, you want to break down the service and provide part of it early, primarily in situations where dissatisfaction is building up. Offering a portion of the service early can prevent this buildup of ill-will (or restore satisfaction) in a situation where dissatisfaction is convex in waiting time. 

So the idea of having people walk out with handheld devices may not reduce customers’ waiting time, but it ensures that not too much dissatisfaction is building up. 

I know that some people love Chick-fil-A and others hate it, but it’s clear that they have a cult following. It’s clear that their employees are usually happier and overall care more about customer service. But what I want to stress here is that Chick-fil-A definitely seems to be making the right choices in terms of queue management, and their management seems to know their stuff. The fact that Starbucks is taking a page from Chick-fil-A’s book, bodes well for Starbucks. 

It was hard ... but I said it!