In the middle of his PowerPoint presentation, Joe Megibow of Expedia Inc. showed an image of a bathroom stall. He turned to his audience and asked how many attendees use their mobile devices here. The question set off giggles around the room, but only a few brave souls raised their hands.
"I do know what the answer to this question is," said Megibow, vice president and general manager of the online travel site Expedia.com, headquartered in Bellevue, Wash. "And it turns out you're all lying to me."
The truth, Megibow said, is closer to 39%, based on a behavioral study from Google Inc. While providing a little comedic relief, his poll question made an important point: If analytics programs want to survive, businesses need a comprehensive enterprise data strategy -- one that enables adapting to a changing environment and also understands how those changes impact a program's data foundation.
Megibow shared his advice last month at the SAS Institute Inc.'s Premier Business Leadership Series. His talk, "Find your Creative Customer Analysis," was full of energy and humor as he walked attendees through specific lessons he's learned over the years. When all lessons are taken together, his presentation gave credibility to, arguably, his most important lesson: The real value of data and business intelligence reports comes when they inspire action. But getting to that last step means making decisions, and that means making mistakes.
This is the second of a two-part series: The first installment deals with the practice of data and customer intelligence; this piece focuses on why a strong enterprise data strategy is vital.
Lesson 5: Evolve or fail
When mobile devices began infiltrating the scene, they pushed businesses like Expedia to broaden their focus. In 2009 alone, 55,000 trips were booked through Expedia's website using a smartphone browser, and the numbers grew quickly. But along with snowballing momentum, handheld devices posed new challenges as developers began creating applications for the emerging medium.
More on enterprise data strategy
Read why a data management strategy without policies spells doom
Learn about the trials and tribulations of implementing a data management strategy
Find out how to bring data management to the C-suite
"Mobile is different," Megibow said. "Applying the same models and techniques you've used for traditional channels to mobile doesn't work."
In building an app, Megibow and his team reconfigured a few basics, such as how to measure things like page views, clicks and conversions. Based on their experience with the website, they also built in customer feedback loops.
TripAssist, Expedia's first app, was released in 2010 and provided an on-the-go itinerary management system. "We launched this app and, literally, right away, we started getting feedback," Megibow said.
In fact, 48 hours after launching TripAssist, customers complained of disappearing itineraries, prompting Megibow and his team to investigate. They discovered that customers with existing accounts who signed in through the app were then greeted by an otherwise blank screen that enabled only one action -- to sign out. Doing so took customers further into the app, but since they were signed out, all of their itineraries were gone.
Lesson 6: Don't expect amazing insights when innovating
Megibow and his team wondered if they could figure out customer habits enough to provide tailored hotel recommendations. They began collecting behavioral data, such as browsing patterns, to do so.
The data provided more dimension so that a map of, say, Florida, went from chronicling where the most popular hotels are located to where the most popular hotels are located for seven different segments. This kind of data is actionable, Megibow said, and could be applied to marketing, targeting and even recommendation algorithms.
The customer segmentations, though, didn't reveal new insights about Florida bookings: Business travelers wanted to be close to the airport and convention centers whereas family vacationers gravitated toward the Magic Kingdom and Universal Studios.
"We started to realize we were looking for the wrong thing," Megibow said. "We [needed] to stop expecting brilliant, amazing insights from this. When we got the models right, they told us exactly what we already knew."
While building models around cities that Megibow and his team knew well didn't uncover unknowns, the prize was realized in how those models could be extrapolated out: The models worked properly, and that meant they could be applied to places the company didn't understand as well.
Lesson 7: Trust the wisdom of crowds; test your way to success
Expedia recently launched a new ad campaign called "Find Yours." When Megibow saw the campaign's new commercial, he spotted a potential problem.
"It was the opening frame: It was just our logo," he said. "In fact, we don't ever say it's Expedia until 57 seconds into a 60-second spot."
Unconvinced that the logo was enough, Megibow asked his team to run a brand experiment: Using an online quiz, test-takers were asked to identify six businesses based on their logos. The pool of businesses included Target Inc., which was correctly identified 99% of the time, as well as Expedia, which didn't fare nearly as well. It was correctly identified just 22% of the time.
"That means four out of five people will basically see this ad and have absolutely no idea who the ad is from until the end," said Megibow, feeling vindicated.
But, unbeknownst to Megibow, his team performed a second test as well. Two different versions of the commercial were shown to test groups: In one, the commercial remained as is; in the other, the company name appeared with the logo. They were identical in every other way.
The outcome was a surprise to Megibow: The original version of the commercial scored better on every single metric by 10% or better.
"Relevancy, recall, likely to buy. You name it, and it scored better," Megibow said. "So I was right, and I was terribly wrong."
Dig deeper on Business intelligence strategy