by Adam T. Sutton, Reporter


CHALLENGE

Mike Brown, VP, Internet Optimization, Vegas.com, and his team noticed that mobile visitors to their destination-based travel and entertainment booking website weren't staying very long.

"People [on mobile phones] who hit the site either were likely to abandon after viewing one page, or they got one or two pages in and said 'Gosh, this is too hard' and abandoned," he says.

The bounce rate of mobile visitors was about 50% higher than that of normal visitors. Their time spent and conversion rates were significantly lower, too.

Had mobile traffic been a negligible percentage of overall site traffic, the team might not have cared. However, about 7% of the site's traffic came from mobile visitors, and it was growing fast.

The team realized it was time to pay more attention to mobile visitors. They wanted to determine whether showing them tailored content would improve their experience, encourage them to stay longer on the site, and increase sales.

CAMPAIGN

Brown and his team built mobile versions of Vegas.com's homepage and specific category pages, such those for gambling, golf, dining, etc.

They then ran a test that served mobile visitors either the mobile-specific pages or the standard webpages, to monitor the difference in performance metrics.

Here are the three steps they took:

Step #1. Limit scope of test to prove concept

The team created mobile versions of the following areas on Vegas.com:
o Homepage
o Category pages
o Hotel room search tool

This essentially gave the team mobile versions of the website's top two layers, which represent the top of its funnel, Brown says.

The team chose these pages for two reasons:

- They could directly impact bounce rates and time spent by immediately severing mobile visitors tailored content.

- Creating these pages was relatively simple, Brown says.

The team did not change Vegas.com's ecommerce architecture, meaning mobile visitors who clicked deeper into the site than a category page, such as to make a reservation, would be using the traditional website's architecture.

Brown recognized this setup would likely cause many mobile visitors to abandon their sessions after reaching deeper pages, hobbling the test's conversion rates.

However, testing the site's ecommerce architecture would require significant investment, and the team wanted to determine whether mobile-specific pages increased visitor engagement before dedicating too many resources to the project.

Step #2. Design pages for mobile visitors

Brown's team created a utilitarian version of its homepage to show mobile visitors (see Creative Samples below). It included:
o Vegas.com logo
o Phone number to call for booking
o Display ad with special offer
o Links to a dozen of Vegas.com's most popular category pages, such as Hotel, Flight+Hotel, Shows and Nightlife

The team designed the page to display cleanly on iPhone, the top mobile device used to access the site. Links were large enough to be easily read and clicked on a touchscreen. The team avoided using too many images to ensure the site would have an efficient loading time.

The category pages the team created were similar to directories. They included links to all relevant information. For example:

- The "Shows" page listed shows playing in town with links to more information.

- The "Hotel" page listed hotels in town, and also hosted a simple search tool to find available hotel rooms.

Once visitors clicked beyond the category pages, they arrived on relevant webpages in the traditional website's format.

"It took about two weeks of time from when we were talking about it to when the test was ready," Brown says. "Honestly, if we had known how easy it was going to be, we would have done it a long time ago."

Step #3. Detect devices, split traffic and monitor results

Brown and the team used a third-party tool to test the pages. They detected which visitors to Vegas.com were using mobile devices and routed them to either the traditional page or the mobile test pages. The team chose a 50-50 split, sending half of all mobile visitors to the traditional homepage, and half to the test page.

The team split the mobile visitors to ensure it could compare performance for the same site traffic. Also, Brown wanted to play it safe. If the test pages were a flop, he didn't want it negatively impact 100% of Vegas.com's mobile traffic.

"When you're testing, 60% of the time you guess right about what's going to work, and 40% of the time you don't, even though when you're wrong it can provide value."


RESULTS

The team proved its hypothesis that visitors on mobile devices needed a tailored web experience.

The team ran the test for just under two months. Comparing mobile traffic on test pages to mobile traffic on the traditional pages, the test pages realized:
o 22% lower bounce rate
o 16% more page views
o 4% higher conversion rate

"People [on mobile phones] are spending more time on the site," Brown says. "They're not as likely to abandon, and people are going to the big product category pages significantly more often."

While the test did "marginally better" in terms of conversion rates, Brown says, it was a result he expected. The team did not test the site's conversion architecture, only the top half of its funnel.

- Prioritizing mobile ecommerce pages

Brown considers the test proof that serving mobile visitors tailored increases results. Since then, his team decided to roll up its sleeves and start customizing pages deeper in the site that are used to book trips and buy tickets. They expect this second-round of tests to boost conversion rates more dramatically.

"We fast tracked the development of mobile booking pages, which were outside this test's scope," Brown says. "Those will be ready in the fall."

Useful links related to this article

Creative Samples from Vegas.com's mobile page tests

Members Library -- Mobile Website Advice from Taco Bell: 5 Considerations to Reach More Mobile Devices

Members Library -- Optimize your Site for Mobile Search: 5 Strategies

Members Library -- Improve the Speed of Mobile Web Campaigns: 4 Factors that Affect Load Times

SiteSpect provided the team's testing platform

Vegas.com