The Rearchitecture of TRIP.com
Darcy DiNucci
When a travel site decides that its business relies on the quality of its customer service, the first step is figuring out just who that customer is.
Travel Web sites are one of the success stories of the Web. Booking flights, rental cars, and hotels online has proven to be one of the most popular e-commerce applications for the Web. TRIP.com got its start in the field early, in 1996, but until recently it kept mostly to its local market, advertising most heavily in its home area of Denver rather than competing head-to-head with the marketing budgets of giants Travelocity and Expedia. In its chosen market, though, TRIP.com was a success, and had built up impressive features, such as flightTRACKER, a feature that let visitors see just where a flight was en route.
In early 2001, the company was getting ready to make a push into new markets. It had recently acquired THOR, a hotel-booking service that offered last-minute bookings and special prices to its customers, and was making plans to create its own reservation-management engine (replacing the one provided by a third party over which TRIP.com had little design control).
"At the time, travel sites were pretty straightforward, you said where you wanted to go, and then sorted through the options," says Jeff Tudor, who was TRIP.com's director of information architecture and design. "Users typically go to three different travel sites before they make the decision to purchase. We wanted to figure out how to create customer loyalty in that environment."
To figure that out, company management began brainstorming a new strategy. The vice president of marketing christened the new directive "customer intimacy," a need to know about and stay close to the target customer through all their travels. To help with that initiative, TRIP.com pulled together a 16-member customer advisory panel - TRIP.com users that fit the business traveler profile that the company most wanted to serve.
TRIP.com knew that to improve its relationship with its customers, it needed to improve its site interface and architecture. "All our development had been system-centric," says Tudor. "We were thinking, 'We have this great technology, now what should we do with it?' rather than 'What do our customers want?'"Looking for help with the redesign, Tudor decided to split the architecture and visual design phases between two different companies. "In film-making, the editor will not be part of the production. The people who are shooting the film are so close that they can't stand back and see a different view," he says. "I wanted someone who could help with the re-architecture and then have a different company come in and do the graphic design." The graphic design for the new site would be done by TRIP.com's in-house staff in association with its advertising agency, Foote Cone & Belding. For the site's architecture and interaction design, Tudor chose the design firm Method. "I wanted a company that would work closely with our in-house staff, so I didn't want one of the large consulting companies that tend to go off behind closed doors and do their work," Tudor says. Tudor liked the fact that Method's principals had worked with those large companies, though, and knew big-company processes.
Fact Finding Methodology Method's first action was to set up a series of interviews with TRIP.com employees. In order to get a handle on the range of requirements for the job, they talked to about 30 people: the CEO and executive staff, plus half a dozen representatives from each department, including travel agents, marketing people, site engineers and others. This is what Method calls the "understanding" phase of the work. "We asked things like 'What's the one thing you would change about this site?' and 'What does "customer intimacy" means to you?'" says Method's project leader and director of user interaction, Ted Booth.
At the end of this two-week process, Method created a document it called the "Re-architecture Program." It echoed back the goals, ideas, and concerns that the Method team uncovered at TRIP.com during their interviews, in order to verify the assumptions that would guide Method through the design phases. The document described Method's understanding of all these points: + TRIP.com's reason for undertaking the re-architecture + The principles that would guide the design - for example, personal attention, respecting boundaries, extraordinary service - and the site's structure - including simplicity and scalability + Success criteria - for example, ease of transacting business and an increase in overall bookings and site traffic + The customer profile, highlighting the "mobile professional" and small and medium-size companies that needed travelservices but didn't have their own in-house travel departments + Guidelines to providing "personal attention", emphasizing the need to imbue every part of the TRIP.com organization with the focus on customer service, and a principle of letting customers always control their experience on the Web site + An outline of the current state of TRIP.com's business + Other initiatives at TRIP.com that Method would need to take into account and weave into their process in order to make the architecture a success - for example, the concurrent implementation of TRIP.com's new booking engine Method also echoed back to TRIP.com some high - level concepts that could guide the project. Those included the idea that the site's service could be thought of as "Travel agent meets concierge" or "AAA plus high-end hotel" - the expected solid infrastructure of information, but with the velvet touch of a luxury service provider. Visual Stories Hearing about TRIP.com's expectations and desires, Method began feeling a hole at the center of its knowledge. TRIP.com was determined to provide the best customer service to business travelers in the industry, but neither Method nor TRIP.com felt they had a model of the business traveler's experience that would help them do that. "Customer intimacy" was the watchword used for TRIP.com's goal, but the team didn't feel intimate with those customers yet. Early in the process, Method and TRIP.com decided to do a deeper round of user research. Method put together a proposal for a round of ethnographic user research based on "visual stories," a method of getting users to document their own ways of doing things - in this case, traveling - so that the designers could craft a service based on those methods, and not on guesswork.
"We needed to get a deeper understanding of our users beyond the common ways we were dissecting them," says Tudor. "We were looking at them from a system-centric point of view: member vs. nonmember, repeat vs. new, transactor vs. nontransactor." With the visual stories, TRIP.com hoped to get an insight into who users were and why users did what they did in their travels and travel bookings. The plan was set. Method proposed a month-long project involving 50 travelers in TRIP.com's target demographic who would be going on a business within the next few weeks. (A recruiting service would help them find travelers matching the profile).
Method sent the travelers out with a kit designed to capture their experiences, plus $125 for their trouble. The kit consisted of a disposable camera and a survey book the traveler could use to record any aspect of the trip that seemed noteworthy, plus an instruction sheet telling each participant how to use them. Each traveler was asked to tell their own story of their experience of business travel. By taking pictures of places, objects and moments, each participant would reveal how he or she perceived the travel experience. What was recorded was up to each participant - the experience would be documented through the traveler's own eyes, not by answers to a set of questions posed by an outsider. For each picture taken, the participant created an entry in the survey book, noting the date, time, place, and observations about their state of mind, the situation, and why they chose to take the picture. After the travel period, 31 of the 50 travelers returned their cameras and books, and Method sat down with TRIP.com to analyze the results. Interpreting the Stories "There were some powerful shots," says Booth. "One guy took a photo of his wife at the house, with an annotation that said, 'Travel makes my wife's life stressful because of the kids.' It was all about the family. He managed to get an earlier flight home to be there for the kids. Another picture was captioned, 'Here's the phone I use to call home to see how the kids are doing.'"
Method began its analysis by laying the 800-plus photos out in piles on the floor of a workroom. "We just looked at them and asked ourselves 'What are these people telling us?" says Booth.
Slowly the group began seeing themes emerge from the stacks. The first grouping identified the types of things being documented: food, what happened in the airplane, at the hotel, and the destination; packing, and getting to and from the airport. The group looked more deeply into what the travelers were telling them with the photos. As the analysis continued, the team merged and split categories. "You find the representative shots - the spokespeople and the 'spokesmoments,' says Booth. The stories told the team things such as: Food is hard to find on the road. Experienced travelers stick to well-worn paths they've carved for themselves at frequently visited cities or airports.
Travelers don't log on the Web while on the road very much. In the end, the team pulled the information into five "big themes": + A Personal Travel System: Frequent travelers develop a personal travel system, encompassing tricks they've learned for making it most efficiently through the gamut of rituals, airports, and cities their trips include + Optimizing Time: Business travelers pay a lot of attention to techniques that will make most of their time on the road + Knowing Your Options: Knowing options for travel methods at the destination, restaurants, and so on gives a sense of control + Bonuses Go a Long Way: Niceties such as upgrades at the airport, good desks in hotel rooms, and other unexpected amenities mean a lot to travelers. Another type of information that grew out of the analysis was an understanding of different types of frequent travelers. Method categorized them in five groups, with different levels of involvement with their destinations: + Minimizer: Focuses on work and not the destination throughout the trip + Tourist Opportunist: Uses business trips to see the world + Home Away from Home: Re-creates the comforts of home wherever he or she is + Reluctant Traveler: Tears himself away from home reluctantly and keeps in touch with family throughout the trip + Budget-Conscious Traveler: Sacrifices comfort for value
As the team sorted through the material, they began to formulate what the information said about the travel experience, and more importantly, what it meant for the Web site. What kinds of features could they build that could help these travelers with the issues they described? Method took its own pass through the material, posted the pictures and their interpretations of them in big white boards, and then invited TRIP.com representatives in to take it to the next step. "We said, 'This is what we found. What do you think is important? What's resonating with you?'" says Booth. "And they would say, 'Oh, yeah. We've been talking about doing this thing on the site about that.'"
From start to finish - getting back the traveler's journals to figuring out the messages they were sending, determining what that all meant for the Web site, putting the information together in a report and in posters for use at TRIP.com - the analysis took almost two months.
The time was well spent, says Tudor. "It gave us a lot of insight into opportunities that we could go out and meet, and the personas were fantastic in terms of helping us define who we were going after," he says. "The combination of the advisory panel, the user stories, and the interviews gave us a rock-solid basis for everything we did. Everything was based on insight, not assumptions. We fought to get enough time to do the research, but when we showed people at TRIP.com what we'd done, we got buy-in for what we were doing throughout the company." Drawing on Experience: Storyboard Workshops at TRIP.com
To continue the site planning, Method once again reached into TRIP.com to take advantage of the company's reservoirs of knowledge and experience with the travel business and the site. This time, the probing took the form of storyboard work sessions with different TRIP.com departments. "We got 20 travel agents in a room, gave them a little storyboard template and said 'Let's plan the reservation process,'" says Booth. The process was repeated with engineers, salespeople, and other groups over two days. Armed with the findings from the visual story analysis, the participants were asked to storyboard three or four different cases, for different types of travelers and different travel situations. "We'd say 'This one is for Martha the Minimizer. She travels every other week. This time she's going from Denver to Houston, to visit a repeat client. It's her third trip there. She has a lot of collateral to carry. She has a corporate travel department and frequent-flyer miles.'" The participants created storyboards in any form they wanted, to any level of detail. Some drew detailed form interfaces, some just the names of screens. All contributed valuable ideas about what the travel site experience could be. "They don't think in flow charts, but they have these great ideas, and they know the business," says Chad Jennings, Method's senior interaction designer.
At the end of each session, Method summarized the ideas in a flow chart drawn on giant Post-it pads, letting everyone see how their individual ideas could be incorporated into a larger site design. Defining the Concept: The Personal Travel Center Method's next task was to take the existing pieces of the site, fold in the new ideas developed in the visual story analysis and storyboard phases, and come up with a model that would tie all the features into a coherent approach. "We usually present two or three different high-level concepts, but in this case, we'd done so much upfront work, it all crystallized in one direction," says Booth. In this case, the team agreed that the site would be thought of as a "personal travel center." The travel center would include streamlined access to the site's basic reservation features, but also offer information and interfaces suited to each customer's particular needs and destinations. And the information and interfaces offered would be the kinds of things that, in the user research, users had said they needed. The travel center would initially feature three ways to access the site's services. The three were a conventional booking form, allowing users to reserve flights, hotels, and cars; a destination centric approach, which pooled all the information about the destination for users to select from; and a past itinerary index, which allowed users to review information and rebook reservations for destinations they often traveled to.
The same services would be available to all users, whether registered or not, but registered users would see richer and more personalized content, based on their past bookings and the information they themselves provided in their member profiles.
The TRIP.com extras, linked to destinations, could include any type of information, from city guides and mapping services provided by content partners to recommendations and ratings provided by other TRIP.com members.Another part of the concept model proposal described issues that weren't spelled out in the conceptual model but should be considered when specifying content and final architecture of the site. Again, gleaned from the user research, they included things like possible structures for interaction with the reservation engine (for example, a Q&A that mimics a traveler's interaction with a travel agent, or the ability to book for multiple travelers with different itineraries on the same screen) and the ability to search for a hotel based on proximity to a business site and a list of amenities.
The document also touched on changes that went much deeper than the Web site. In order to create customer intimacy, it proposed features such as individual agents assigned to particular customers, emphasizing the degree to which the Web site was part of a larger shift toward a customer-focused business. Paper Prototypes The next stage in Method's procedure was to flesh out their plan by building and testing prototypes of the proposed pages and interaction sequences. In the first phase, Method created page models on paper, with just enough detail to test the architectural concept and some ideas for the interface elements and features with some sample users.
The prototypes Method created for this project were "plain vanilla," in Booth's description. "We do them in black and gray. At this stage, it's not about visual design. It's about what you call things and what's on the pages."
Interestingly, this architecture project included none of the large tree diagrams so familiar in Web design. "We don't even do the big architecture diagrams anymore. Clients just don't understand them," says Booth. "We've found that moving straight to prototypes is the way to go." As the prototypes developed, Method pinned the pages onto large foamcore boards in a war room, moving pages around, adding Post-its, and inviting other firm members into the room to discuss possibilities.The first round of user tests involved 16 existing TRIP.com users and others fitting the target demographic, who were asked to try to book travel for their next trip using the features shown on the paper prototypes. "They'd say things like, 'I fly to Chicago a lot. I wish there was a way I could search all the airports in Chicago at once.' When something like that happens, we have someone there who can draw that feature up on a piece of paper and show it to them right there." As the tester and the note-taker sat in the room with the test participant, members of the TRIP.com staff remained on the other side of a one-way mirror. In that room, all the paper prototypes were posted on the wall; the observers could stick Post-it notes to the pages as issues and ideas emerged from the tests. After the day's session, Method and TRIP.com would regroup and summarize what they'd learned.
"During and after the day's tests, we'd notice what was and wasn't working, come up with new ideas, and test those the next day. Over four days of testing, we essentially went through four different versions of the architecture," says Booth. Now in HTML Now it was time to build the site in HTML. "We'd build screens for three days, show them to TRIP.com for feedback, and refine them for another three days. Then we'd do more one-on-one sessions with users," says Booth. "At this point, the high-level structure is in place, and you're dealing with high-level interface issues - what are the contents of a drop down menu, should this be a pop-up calendar, that sort of thing."
One interesting thing the team found was that their efforts to keep pages streamlined and minimal (a general Web doctrine) were hard to stick to. "The pages really bulked up between the paper prototypes and the final HTML prototypes," says Booth. "We found out that you just had to have this feature or that information there." They also found that their prejudice for simple, focused pages could be open to question. "We learned that information density is not a good or a bad thing. It's how you package it and how relevant it is," says Booth. "We kept things pretty simple during the reservation process. But when they finished that, at the guide phase, people would scroll down and get all sorts of information about the trip they just booked - weather, flights around the same time, gate layout at airport - and they liked it. It was relevant to their trip and it was a point when they were open to it."
The End of Architecture The last piece Method delivered to TRIP.com was a set of specs that documented each feature and the rationale for each page proposed for the new site. The notebook they delivered included a CD-ROM that held the clickable prototype, plus an annotated print out of each page. An introductory section summarized the strategy, concept, and principles on which the site was built.
"Engineers can take the HTML pages and just hook their code up to it. Designers can take it and just move things around," says Booth. "The engineers loved it because it explained how we did things." Working with, Not For The detail of Method's spec was based on the same principle as the rest of their work with TRIP.com. In the industry, it goes by the name "knowledge transfer" - making sure that the thinking that created the plan was transferred into the company and doesn't leave with the consultants. The way Method and TRIP.com worked together was carefully planned from the beginning to make sure that the TRIP.com staff was included in plans and aware of the reasons for and thinking behind each step. "In the end, all we were giving them was and architecture, but it's up to them to build it and keep it up over time. You have to get their ideas and their expectations, and they also have to have some ownership," says Booth.
"I told my staff that these people are going to come in and help us for a while. Our company was the technology and we knew it inside and out," says Tudor.
Booth also points out the importance of each department's involvement in the thinking as it developed. "Here's a company that's built entirely around this Web site, but they each have their own departmental point of view. We wanted to develop a coherent structure they could all work around," he says. They also found that their prejudice for simple, focused pages could be open to question. "We learned that information density is not a good or a bad thing. It's how you package it and how relevant it is," says Booth. "We kept things pretty simple during the reservation process. But when they finished that, at the guide phase, people would scroll down and get all sorts of information about the trip they just booked - weather, flights around the same time, gate layout at airport - and they liked it. It was relevant to their trip and it was a point when they were open to it." Keeping it Honest Much of the work that Method and TRIP.com completed has yet to be implemented on the site. The plan spelled out in the specs was understood to be a long-term vision that could only be accomplished once the work of seeking out content partners and major engineering projects was completed. In addition, in the middle of the redesign process, TRIP.com was bought by Galileo (who, with Sabre, is a leader in providing scheduling information to travel agents), and as often happens on the Web, fast changing business requirements have necessitated a new strategy. The specs remain, though, to spell out the reasoning behind each design, to be implemented as the opportunity allows. At press time, the site had implemented the basic navigation plans, the personal travel center, and the reservation process. Tudor is immensely proud of the thinking the team accomplished, and especially of the insights they arrived as through their investment in understanding the site's users. "The value of user-centered design is that it keeps you honest. As you become immersed in an industry, you can start to make too many assumptions about who the users are and what the opportunities are," he says. "Some people might think the obvious way to do a travel site is to reverse-engineer Expedia and go," he says. "But you do that, and you will always be a day late. There are opportunities that are missed because you're a follower and not a leader."
