Collette Vacations had a powerful marketing tool at its fingertips: A database of customer satisfaction survey responses, which also held information about consumers' travel plans.
There was just one small problem: The questionnaire was designed by the tour packager's product development staff, and customer ID numbers hadn't been considered an important data point — so they weren't recorded. None of the data could be tracked back to an individual client.
The ability to solicit past customers was essential to Collette's success. Despite reasonably high satisfaction levels, the return customer business — people taking another Collette tour within three years — was running some 40% lower than its competitors.
Being able to address concerns and desires on an individual basis was a big part of Collette's win-back strategy. Next-likely-destination information was useful for designing new programs, but knowing exactly which customers were interested in the various trips would have been more so.
In truth, there were more problems than that with the survey. It was geared toward monitoring a trip's services, and many of the questions (such as cleanliness of the motor coach or whether the tour guide wore the official uniform) were not, as Collette's direct marketing manager Diane Gorine puts it, “model-able.”
Uniting survey data with customers proved, if not easy, at least possible: The company had kept the original surveys, and was able to back-match the responses. Pawtucket, RI-based Collette augmented this information with customer demographics, including age and location, and transaction data such as the number of trips taken, destination and price. To give the file extra strength, it added marketing survey information regarding potential travel plans and interests.
Gorine then evaluated the survey questions, hoping to striking a balance between those needed for quality control (“Did your tour guide wear his/her uniform at all times?”) and those that would provide insight for the marketing department (“What is your next likely tour destination?”).
She notes that relying on survey data, even if the right questions are asked, is risky. If only, for example, 30% of the questionnaires presented are returned, chances are the responses are from only the most satisfied, or the most dissatisfied, customers.
To get around this, Collette asks travelers to fill out a satisfaction survey the night before the tour ends. This aids in completion rates, which are around 86%, Gorine says. It also reduces the likelihood that a bad airline experience — a factor over which Collette has no control — will color a traveler's opinion.
These responses, along with the other data, are plugged into Affinium Model, a predictive response tool from Unica Corp. Gorine herself runs queries from her desktop — a setup that lets her test a wide variety of variables without having to burden an IT department.
What has helped Collette build its current predictive file is the completeness of each of these sources. Demographic data is collected during the booking process. The marketing survey is sent to customers and prospects on the company's file, and is used to design future offerings.
W
For more CRM and database marketing material, go to http://directmag.com/disciplines/crm/.
Bear in Mind…
Direct marketers planning to add survey information to their databases should keep the following concerns in mind during the poll design and data review phases, says Collette Vacations' direct marketing manager Diane Gorine:
- Who will be the primary end user of this data? Marketing and customer service executives will ask different questions.
- Will the questions be useful for predictive modeling? Non-quantifiable questions will have limited value, and responses separated from respondents won't allow a DMer to base customized solicitations on the analysis.
- What's the survey's response rate? And what might an overly high (or low) response rate mean? Low response could indicate that only very satisfied — or very dissatisfied — individuals are answering. This will skew a modeler's insights.
- Does the response rate vary by region or tour? Responses could be different depending on the product. As Gorine notes, there's a world of difference between Australia and Austria.
- How fresh is the data? Twenty-year-old data will yield 20-year-old insights. Collette Vacations limits its analysis to travel done within the last three years.
- How is the data stored? Not having customer IDs linked to survey data was a problem for Collette. But it would've been far more difficult if it didn't have the customer IDs at all.
- Are responses related to contact data? If the information is going to be used for customized solicitations, it's essential to know which customers or prospects submitted particular responses.
-
How should multiple survey records for a single customer be treated? Collette runs different analysis based on the last tour taken vs. all tours taken, and compares campaign results for each to determine predictability.
— RHL
