Dara Khosrowshahi

INTERVIEW: Expedia Inc. CEO Dara Khosrowshahi

"Fail fast" is a common mantra for technologists and Silicon Valley entrepreneurs. It begs us to experiment, to learn lessons from failure and to quickly adapt them toward something that works. In a keynote address on innovation at The Beat Live in Chicago last week, Expedia CEO Dara Khosrowshahi added an appendix: "Fail fast; fail small," he said. He discussed how the pace of innovation is accelerating, how experimentation and collaboration nurture innovation and how "perfect is the enemy of innovation." Along with that, he shared a philosophy on the virtue of making "small bets" in various areas versus making "big bets" in one or two. Following his address, Khosrowshahi took the stage beside The Beat editor-in-chief Jay Boehmer to field questions on innovation, the corporate travel market and Expedia's bets__big and small__in travel. An edited transcript follows.

Boehmer: Are there any emerging innovations or technologies that are worth a big bet right now?

Khosrowshahi: The pattern that we have is: We do make small bets and then a big bet for us is 20 small bets around something. One of the disciplines that we have around our engineering is: We're constantly trying to think about what we call an MVP. What is a minimum viable product? A lot of times, a big bet is building this giant thing that's going to take a year to build, and we've got a plan for it, et cetera. The best engineers are cheaters. What they do is figure out how to cheat so that instead of building this thing that's going to take a year, I can drive a shortcut that will allow me to build it in two weeks or a month.

This is a little bit of a secret; don't tell too many people: If there's a feature that's particularly complicated to build, we'll build the user interface and we'll say, "Hey, if you're interested, click here," and when the user clicks there, we tell them, "Oh, it's broken." But the thing is we never built it, but we want to see if users were interested in it. Then, if you see a lot of users clicking on this thing, you're like, "Hey, I've got some data back that this thing that's going to cost us six months of engineering time is worth it to go out and actually build it." Don't tell that to too many users. Sometimes, things really are broken, but sometimes we ... That's what we call smoke test.

One space that we're certainly making bets around is customization and in-context customization. This is especially useful on the corporate travel space. The issue that we have on the leisure travel space is that most of our customers don't travel that much; they're not frequent travelers. We can, on average, identify about 20 percent or 25 percent of our travelers, and then everyone else is a stranger to us.

As more and more people sign in with Facebook, as interactions happen over Messenger, those interactions come over platforms that have identity attached. We're seeing the percentage of consumers that have their identity attached go up and up and up. With corporate travel, you have 100 percent identity attached.

Once you're able to identify that traveler, you can start customizing experiences in really, really cool ways. On the corporate side, we know your offices are in Chicago. We know what flights you like to take from Seattle to Chicago, and we can offer up those flights to you in a really easy way, so it's literally: Here's my flight, click, done. This is something that we're learning from the corporate space and want to take it to the leisure space.

Then, I do think that artificial intelligence is another overall theme that's happening. Artificial intelligence especially around customer service in areas where you've got identity attached is something that's super interesting for us.

Boehmer: How long before AI matures?

Khosrowshahi: It's going to take some time. There are two flavors that people talk about. One is the sexy flavor and one is kind of unsexy. The sexy stuff is being able to have a conversation with a robot and booking travel, et cetera. Especially over messaging, I don't have to talk to anybody. Those interactions are much more difficult because you need huge amounts of data in order to drive strong AI behavior. You need very repeatable behavior in order to drive it, and the more wide the interaction path is, the tougher it becomes for you to program an AI that works all the time. The issue with users is that if they get a wrong answer, they say, "This is waste of my time, forget it." I've used Siri, and Siri is pretty good, but after two or three times of Siri not understanding, I'm like, "Forget it; I'll just type it. I don't need this."

Conversational AI is the sexy stuff, and it's going to take a while. It's going to be five to seven years before it becomes great. I think it will be the Googles and the Apples of the world and Amazon that perfect that stuff.

Then, there's AI around very repeatable tasks. If you message us, "Cancel my flight to Chicago on Wednesday," that is not something that a customer service agent is adding a bunch of value. That's something that a bot can easily take over, and it can be an instant interaction and can be delightful.

Boehmer: Egencia is viewed as a good fit for smaller, midmarket companies. What is your sweet spot for Egencia? Is that evolving?

Khosrowshahi: It's absolutely going to evolve, and I think most Internet technologies that have really taken off over a period of time start with the broader sector, small businesses, and then they move up. That was where Egencia was seven, eight years ago. It was more a small business type of application, and we kept adding to the Egencia platform more and more and more where now we've got P&G using Egencia.

As the platform has gotten more powerful, as we have gotten better on the service side and as some larger clients have been willing to compromise in order to drive more online adoption, we have gotten to a sweet spot where we are able to now serve multinational customers who are spending in the hundreds of millions of dollars. But it does require some compromise in that there are custom builds that some larger companies have had historically. The ones who are married to the custom build typically are not the right ones for Egencia. The issue with custom builds is that they're a dead end. You need scale, you need data in order to continuously improve. If you build a custom build, it's really cool, but then five years from now it is totally dated and you've got to rebuild it. That's not Egencia's bread and butter. That's business better done by other players. That's the biggest factor now as to whether we can or want to serve a big multinational. If they want a huge amount of customization, there are other players who can help them out better. If they want a platform that's improving 10 percent a year__in other words, it's not about what the product is now, you're making a bet on the product a year from now and a year from then and a year forward__then we think Egencia is a pretty good bet. There are some big players who are making that bet.

Boehmer: Who does Egencia view as its biggest competitor: Is it the travel management companies or is it Concur?

Khosrowshahi: Rob Greyber, Egencia's president, looks at all of them. Egencia is one of the most paranoid of our family, so they're constantly looking at a bunch of threats. Because Egencia views itself as a service company and a technology company, they tend to look at technology companies like Concur and, let's say, the Expedias of the world, as competition. But at the same time, they also look at the [American Express] GBTs of the world and the Carlsons of the world and these businesses that have an incredible base of business and relationships with clients that they've had for 20 or 30 years. They look at them and say, ultimately, that's what we want to be, but I want to be a version of that that is super technology powered.

Boehmer: On earnings calls and other forums, you've alluded to further M&A in the corporate space in particular.

Khosrowshahi: More than allude.

Boehmer: So, what are you going to do?

Khosrowshahi: I won't get into specifics, but one of the tough things about the corporate space is: You've got a customer base that is very sticky. The leisure customer base is not very loyal, but the corporate customer is very sticky. And so Egencia's growth is based on our building up these sales teams and account management teams, and there's a structural limit to that kind of growth. You don't want to overstretch that kind of growth. You don't want to sign up too many clients so that when we bring them on, you have difficulties.

There's a certain growth rate that's about right for Egencia, and it's 10 percent to 15 percent a year. And they keep organically gaining share against an industry that's been largely flat, and that's great. What we absolutely will do then is complement that organic growth with acquisitions over the next five to 10 years, so we can get to real big scale at Egencia. In the next five years, my guess is we will have two to three acquisitions with Egencia.

I think they'll tend to be in the smaller and midsize bucket, and there will be great companies that have terrific service and a great client base who we believe will make the transition to the higher tech, higher percentage of online transactions that Egencia brings while also properly servicing the stuff that needs to be serviced. I just don't think that an operator is adding a bunch of value booking a flight from Seattle to Chicago. That stuff should go online, and there are some clients who are happy to go online and so on.

Boehmer: When you look at M&A, are you looking here or abroad?

Khosrowshahi: Everywhere. I mean, we want to increase Egencia's global footprint so that the big global players can work with us with a single platform. We have partners in Brazil, et cetera, but it's not quite as tightly integrated as, let's say, the Egencia platform. To the extent that we can build our global footprint in strategic markets, we will do that. That comes with a little more pain as far as then driving online adoption, et cetera, but we will make those kinds of investments, and we'll look to scale where we have an absolutely spectacular product. So, in the U.S., we have, I believe, a spectacular product and why not just bring more volume onto that product? It's a lot easier. I think we're going to actually drive both.

Seapark Consulting consultant Margaret Brady: As a competitor, Concur has online and expense. When you look at the full end-to-end lifecycle of travel, expense is such an important component. What do you see as your play, if any, in connecting with expense systems or creating your own solution?

Khosrowshahi: When we look at Concur, there's this idea of coopetition. We compete because they've got a travel solution, but we cooperate because we actually line up with their expense solution pretty well, and we're integrating our travel solution into their expense solution in a way that's much stronger now than it was three years ago.

We don't want to be in the expense business. We want to be in the travel business. Our approach is going to be to integrate with the expense providers out there, and Concur is a big one in the U.S. And then there are a lot of other players outside the U.S., as well. We do have our own play in Traveldoo, which is a French company that is in the booking tool business and the expense side, and we are building that up over a period of time.

Whether that turns into something big or not remains to be seen, so we do have a little bit of a footprint on the expense side and call it a small bet that is starting to get bigger. It's a really, really good product, but it's a European product, and we'd like to make it more of a global product.

Boehmer: We've seen a bunch of hotel direct booking initiatives come out this year. Where are we now? Are hotels moving the dial at all with these initiatives? They clearly have Expedia in their sights.

Khosrowshahi: It remains to be seen. Any company out there, anyone who's distributing on the Web, is going to want to have a balanced distribution portfolio. We go out there and we're advertising on Google and on metasearch, et cetera, and at the same time we don't want to be too dependent on third parties for our distribution. We want to have the majority of folks come to us and book direct, and they do. That's a sign of a healthy business.

Some of the hotel players are trying to do the same thing. The issue, and the interesting problem as it relates to some of the hotel chains, is that the newer generation of travelers__certainly the leisure traveler__seems to be much less hotel-brand loyal. They want authentic product. They want different product. They want to be special.

What we've seen as far as Expedia users: If you look at, the percentage of people who actually say "Show me a hotel from the top hotel chain" is less than 1.6 percent of searches. These are people who are not brand loyal.

To some extent, with the hotel direct booking initiative, they are appealing to a segment that they already appeal to, who is really interested in joining hotel loyalty programs, et cetera. I personally feel that they're just stealing loyalty members from each other.

There's this new generation that they can reach out to who hasn't driven their loyalties yet. And they can go out and acquire those customers, and with great service and data and customization, they should be able to convert those users into loyal users. So there are certain partners like Red Lion where we are actively helping them merchandise and acquire loyalty members. If the member then comes to Red Lion the second time direct, then we've done a great job. Over a period of time, the economics will take care of itself, so I'm hoping that we get to a place with some of our hotel partners where the data interchange between us and them is much more rich and we can actually be a very, very strong technology provider to them and we can be a terrific customer-acquisition vehicle for them. That's our goal because clearly it's something that they want.

Boehmer: What is the state of hotel relationships right now? They seem fraught.

Khosrowshahi: In general they are some good, some not great. I think that the hotel chains and certainly the owners are worried about how much, how fast the audience is moving to the [online travel agencies]. I think they are trying to figure out their strategy there. There is either a try-to-beat-them or a join-them strategy. I personally think that we can be an incredible customer-acquisition vehicle for them, and it's my job to convince our partners that we can. Obviously, we haven't yet done a good enough job.