• May 1, 2009
  • By Melanie Polkosky Human Factors Psychologist & Consultant - IBM/Center for Multimedia Arts (University of Memphis)
  • Interact

What I Learned Waiting Tables

Article Featured Image

Sometimes it flashes through my head that everything I need to know about voice experience design I learned from waiting tables. OK, maybe not everything, but  knowing your survival depends on a good tip can really crystallize how to behave when you’re taking care of a customer.  

What if everyone who had something to do with automated customer service had to wait tables? In deference to a popular deli in my hometown, I’ll refer to our hypothetical server as “Tron,” which is truly what we wait staffers were called back in the day.

Tron: Hi. Welcome to Downhome Eats. How can I help you?
Customer: Um, I guess I’d like a table.
Tron: OK, what drink do you want to order?
Customer: I’ll take a soda.
Tron: I didn’t get that. Please select root beer, lemonade, or milk.
Customer: Seriously? Those are my choices? Don’tcha have Coke or something?  
Tron: Coke. OK, what meal would you like?
Customer: Can I just think about it for a minute, please? I haven’t even seen a menu.
Tron: I didn’t understand that. What food do you want to eat?
[The door swings assertively as the customer stomps out.]

Oh, sure! Go ahead and laugh! In a slightly different context, what is frequently touted as the way to design prompts sounds pretty silly. It’s a clear violation of our expectations about appropriate customer-service behavior. 

In this interaction, Tron violated the “restaurant script” in several ways: She wasn’t helpful and polite, didn’t anticipate the customer’s needs, and didn’t perform actions according to the customer’s expected sequence (give a menu before taking an order). The resulting dialogue is stilted, leading to a negative customer outcome.  

As any server worth her tip knows, the language of the interaction clearly demonstrates a relationship between Tron’s behavior and the outcome.

Now consider a more expectation-consistent interaction:

Tron: Hi. Welcome to Downhome Eats. Just one for dinner?
Customer: Yep.
Tron: OK, here’s a menu. Our special tonight is baked lasagna. I recommend it. What would ya like to drink?
Customer: Coke.
Tron: Thanks. I’ll be right back. [She leaves and returns with a Coke.] Here ya go. Are you ready to order?
Customer: Yeah, I’ll take the lasagna.
Tron: Great choice. What dressing for your salad?
Customer: Thousand Island.
Tron: OK, I’ll get this order in and be back with your salad and some bread. [She walks to kitchen.]

This example doesn’t seem as contrived; it uses a typical sequence of events we expect when dining out. It’s friendly, familiar, and polite, and guides the customer’s choices while incorporating some personal touches (dinner recommendation and positive reinforcement for the selection). According to objective metrics, the first interaction should be better because it’s more efficient (59 total syllables versus 78, 11.8 syllables per turn versus 15.6). But these measures don’t capture the critical subjective difference: The tone and formality of Tron’s conversation are responsible for differences in error frequency and outcome.

So in honor of that long-ago, joyous time of youthful table-waiting, let me encapsulate my vast restaurant-based learning into one single sentence: Always be as agreeable and pleasantly unobtrusive as possible.  

A pocket of researchers focus on the relationship between wait staff and the tip as an indicator of customer satisfaction. As with voice interaction, outcomes depend on factors beyond the server’s control, such as weather, background music, size of party, customer gender and mood, amount of alcohol consumed, and payment type. Tipping also depends on the server’s behavior, such as smiling, light shoulder touches, introductions, drawing doodles on a check, and being polite and complimentary. 

This research is particularly relevant for voice interaction designers too. The social-communicative and nonverbal behaviors that servers control are the same ones that designers control. Research has shown that individuals can identify  whether the speaker is smiling only from hearing a voice. Impression management theory applies to the designer’s job: We are tasked with shaping vocal and linguistic behaviors to create the best possible impression of an automated speaker, which creates desirable business outcomes.  

A recent cartoon opined that businesses might finally have to resort to good customer service in our challenging economic environment. Ask yourself the hard question: How big a tip would you leave your IVR? If the number is surprisingly low, I’d humbly suggest you wait tables. You’d be amazed what you can learn about human behavior.

Melanie Polkosky, Ph.D., is a social-cognitive psychologist and speech language pathologist who has researched and designed speech, graphic, and multimedia user experiences for more than 12 years. She is currently a human factors psychologist and senior consultant at IBM. She can be reached at polkosky@comcast.com.

SpeechTek Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues