Applying Decision Trees for Value-Based Customer Relations Management: Predicting Airline Customers’ Future Values

Explanation of the Conclusions Reached By the Case 

Tirenni, Kaiser and Herrmann (2007), discusses a case study of an airline that uses the decision tree in the segmentation of customers into four categories and predicting their value. They include long term, middle term, new customers, and prospects. The decision tree is applied to each group to predict their customer values. From the conclusion, it is clear that gender, previous spending, email address, age, choice of seat, nationality, and whether one had started a family or not determined their customer value. However, a review of more than twelve months narrowed down to the previous spending eliminating most of the other variables. For instance, low-value customers and prospects mostly chose to receive their email at home, didn't choose a seat, were young without a family, female, and from certain countries. The high-value customers value on the other hand, as business oriented and cooperate. They show high previous spending under the periods of review, they are likely to choose their seats and prefer aisle seats. In summary, it indicates that the use of customers' previous spending behavior predicted the value of the customers. However, where the information is limited as the case for future, new and prospect customers, demographics and behavior were the most probable way of prediction. 

How to Improve the Model Proposed For the Case and Other Variables to Consider 

The model could be improved by adding two more variables that mostly predict the passenger behavior then performing a t-test against a test group. They include the duration of the booking before departure and the time of the year the customer or the prospects prefer to travel. The business passengers were most likely to be of high value based on the models finding. Therefore, the model could go a step further and determine their estimated duration of the booking before departure and the time of the year they fly. Both variables determine the value the airline gets out of the customers. Customers that book very close to the departure date most likely pay higher. Similarly, those that do not discriminate the seasons also are likely to fly even during peak seasons when the fares are high and during low seasons when the airlines have less passenger still adding value to the airline. After obtaining the results, they should be subjected to a t-test against a test group preferable of known characteristics. 

References 

Tirenni, G., Kaiser, C., & Herrmann, A. (2007). Applying decision trees for value-based customer relations management: Predicting airline customers' future values.  Journal Of Database Marketing & Customer Strategy Management ,  14 (2), 130-142. DOI: 10.1057/palgrave.dbm.3250044