United Airlines: Dude Use Your Data to Solve Your People Problem

So it’s been less than a week since United Airlines has been dragged through the media because it basically ordered a paying customer dragged off one of its flights.

The ocean of mea culpa’s by United’s CEO Oscar Munoz hasn’t even begun to stem the ongoing tide of haterade pouring into this drama.

The public relations nightmare began almost immediately, with a dramatic recording of the event uploaded to Twitter, a bloodied, battered and bewildered victim seem wailing in righteous opposition to being removed, a billion drop in stock value and the denouement — lawyers and angry family members at a press conference in Chicago.

United seems to be the whipping boy of all things airlines but it’s figuratively becoming the worse airline in history; winning the race to the bottom with its unbearable customer service.

And while some folks may see this as a people problem at United, I see this entire incident as a prime example where data and technology, not people, could have solved this a lot more eloquently with a lot less pain.

Human Computer Interaction Gone Wrong

So I’m a design researcher. Before that I was a journalist. So I’ve spent the better part of the last 30 years observing people and how they interact with the world. I specialize in how humans interact with technology and work on designing systems that improve that relationship.

The horrific images of Dr. David Dao being dragged away from a product that he rightfully purchased showcases in technicolor a human-computer interaction gone horribly wrong.

Make no mistake, this entire incident is about people. It’s about the passenger who enter into a service contract with the people who work at a corporation who provide that service. That’s the nuts and bolts. But the reality is a bit more sticky.

The airline is engaging in a bit of game theory here. Yeah, the game theory from A Beautiful Mind, that’s the one. You can read whole books about game theory but for layman’s understanding is all about calculating the odds of human decision making (behavior).

In the United incident, you can also call it playing a game of chicken.

United, and most all other airlines often over book flights. Using mathematical models they calculate, and mostly rightly so, the probability of every passenger showing up to take all the seats sold. A lot of times not everyone shows up and the airline gets to sell that seat again or use those seats for a practice known as “deadheading.”

In this case, United wanted to deadhead four employees, fly them from Chicago to Louisville for free. The flight wasn’t oversold but everyone did show up for their purchased seat. This put United in a sticky quandary.

How to get four extra, empty seats where there are none. Since you can’t make something out of nothing (thanks Einstein), they asked for passengers already seated on UA3411 to volunteer their seats.

Losing the Game at the Start

Putting people on the plane and then making a request for volunteers was the first of many mistakes United made that day. This was illustrative of another theory on human decision making called prospect theory.

As Kevin Zollman, a philosophy professor at Carnegie Mellon rightly pointed out in this NPR piece, people don’t like to give up what they have. “You tend to value something more once you have it,” Zollman said.

He’s absolutely right.

Prospect theory is all about how we calculate economic risk. When airlines try to get passengers to give up their purchased seats they’re trying to get a person with a bird in the hand to give it up for two in a bush. This proposition works a lot better if the person asked to give something up isn’t already sitting in the bush.

In United’s case people were already seated on the plane when the proposition was offered. That was their first mistake. Prospect theory says people don’t care about winning as much as they care about not losing.

Taking something away from someone when they have it, is a lot more difficult, even if you’re offering them something that is seemingly more valuable than what they already have.

I see this in design…or more accurately redesigns…all the time. Take Craigslist.org, for example. Since the online marketplace debut in 1995 there’s been a cry for it to be redesigned and repurposed. Designers everywhere bemoan it’s ugly user interface and horrible design. And start-ups salivate at it’s general lack of customer interaction or engagement seeing an opportunity to provide better service. Yet Craigslist remains the 15th most popular website in the U.S. It’s design hasn’t changed since the late 1990s, yet it still rakes in about $300 million annually.

Craigslist should have gone by the wayside by now but it still looms large. It’s good enough. And sometimes that’s all you need.

Back to the airplane seat. United asked for volunteers to give up a seat they already were in and change a trip they had basically already started. They offered $400 but it wasn’t enough. Then they offered $800. And still no takers. But really they were on the losing side of game theory the minute they let the people on the plane.

Once you let the passenger on the plane the airline gave up any leverage in the game. It was inevitable a confrontation would ensue. To avoid that confrontation United tried to use math again, and again, they misfired. Using a mathematical formula that takes into account loyalty programs, flight-check in times, ticket price, etc., the airline chose passengers “at random,” to be “re-accommodated.”

This may seem like a rational solution but as you witnessed, human behavior often is far from rational. There was no contingency if a passenger resisted being “re-accommodated.”

A Data-Inspired Humane Solution

As a person who spends much of her time helping to design data-inspired solution I immediately thought of a way that well-chosen data and good human instincts with a better understanding of game theory could have provided a non-violent solution. With the caveat that the real answer is to completely overhaul United’s computer systems to prevent such a mathematical fiasco let’s look at an alternative way of solving United’s Sunday afternoon problem.

Once agents realized they needed four seats for deadheading employees here’s what could have happened. While the passengers are in the boarding area, United could

  1. Execute a simple search on the passenger manifest and cross-reference that search with the passenger travel history to come up with a targeted list of names of people who had voluntarily given up their seat on a plane before. If none found, calculate the probability of which passengers would give of their seats using a combination of biographical, economical, attitudinal and behavioral data points.
  2. Once the list of names possible volunteers had been established; calculate the average price of ticket bought to come up with an apex value, a value more than the average ticket price but less than the $1,300 or so bucks an airline can offer to volunteers.
  3. Zollman, who has written extensively on game theory, suggests the request should not be done publicly, rather privately out of earshot of others. You can read why here. I suggest making the offer as soon as the logistical problem becomes evident. Airlines can contact passengers in a number of ways…text, email, calling them up to the ticket counter. However, they chose to do it, United could make the request of volunteers to a small targeted audience quickly and quietly. This would have raised the probability of success.
  4. If no offer was taken by the targeted list of volunteers, United could have done a reverse auction, starting at the highest ticket value and going down until someone swooped it up. Zollman says doing it this way makes people get engaged as if they’re part of a game. And I agree. Now the ticket value seems like a win; instead of a loss because they would have been competing against other passengers for the best deal.

The key is to make the decision seem like the person is avoiding loss. This is an essential lesson learned from prospect theory. People like winning. But they like not losing more. Our aversion to loss is much, much stronger than our will to win.

In reality, United set up all its passengers to lose by putting them on the plane. Losing a seat is more risky and, ultimately hurtful, than “winning,” money. Technology can be programmed to understand this even if humans don’t.

None of the above actions could have guaranteed that four volunteers would have been found but they seem a bit more reasonable and probably but a humane solution would have been created.

Either way, I do believe that it is a case where technology and humans working together could have created a better outcome.

What do you think?