• by Stephanie Miller , April 20, 2010

Our opening keynote speaker this morning leaned over to me at breakfast and said, "Who puts the database guy as the opener?!" The good news is that we got to listen to Ted Wham, VP, Database Marketing of Orbitz. He's talented, experienced, fabulously geeky and passionate about using data intelligently to improve the customer experience.

Ted generously walked us through how the Orbitz business works and some of the unique challenges of being an online travel agency, like the fact that they don't own the inventory and do not control pricing (or know it in advance). Despite the huge database built over an eight-year history, the only customers they email are those that meet the dual requirements of opt-in subscription (permission) and deliverable (non-bounce).

After a detailed list cleansing and data enhancement project, Orbitz had a very clean file that Wham and his team enhanced with a long transaction and click history. Each link has a unique name and tracker for reporting and future customization. Wham recommends that you name your links with real English names so that you can easily manage them. Orbitz uses seven categories like AIR, HOT, CAR etc. This relates back to the high number of customizable elements for each email message. A "simple" hotel listing, for example, includes some combination of dozens of elements like amenities, price and location. For email marketing, these listings can be customized to the subscriber history and recent behavior. Combined with behavior, Orbitz aims to have a limited number of clusters (e.g.: young families or empty nesters) where copy treatments can be targeted.

Use of this intense customization capacity in part resulted from bulking by one of the major ISPs. With high frequency typical in the travel industry, Orbitz was having some trouble with inbox deliverability. Segmentation is a great way to improve relevancy and lower complaints (clicks on the Report Spam button), which is the biggest factor of sender reputation and inbox placement.

A traditional RFM (Recency, Frequency, Monetary Value) model was not deep enough to segment the file or customize the content -- and keep spam complaints low. So Orbitz refactored the analysis by adding engagement measures specific to email marketing - like recency of email click or frequency of email open. This allowed much more intelligent segmentation. It is a bit counter-intuitive to lower weight on "best buyers" (traditional RFM) and more weight on "good buyers" (those who were active with email marketing).

"Hey, but if you don't reach the inbox, you can't get a response from any buyer," Wham says.

Reminding us that email marketing IS database marketing, Wham suggests, "Grab that data and use it to improve customer communications, and response."