Publication Date
2016
Document Type
Dissertation
Committee Members
Mary Fendley (Committee Member), James Hamister (Committee Member), Nan Kong (Committee Member), Yan Liu (Committee Member), Pratik Parikh (Committee Member), Xinhui Zhang (Advisor)
Degree Name
Doctor of Philosophy (PhD)
Abstract
Jewelry retail companies like Fred Meyer Jewelers employ direct marketing such as mailing lists and catalogs as a primary means to promote products, maintain customer relationships, and drive sale growths. Most jewelry customers, however, do not purchase frequently – they make only a limited number of purchases at special times with long lead-times between purchases. This has created a series of challenging problems for the marketing team at these jewelry stores in a) predicting when and what the customer purchases, and b) deciding what customers to be targeted so as to maximize the return on investment in the design of market campaigns. This research aims to provide data analytics and mathematical models to the above market campaign optimization problem. The methodology is particularly designed to uncover the special times or events such as birthdays and anniversaries, which are unique to each customer, yet not available to the jewelry retailers. In a nut shell, the methodology is composed of several steps as follows. First, to mine these hidden intrinsic attributes, a time-based sequential pattern model is developed to discover customer shopping patterns (rules) and the lead-times between left-hand and right-hand purchases in these patterns. Second, innovative analytics have been developed to calculate each customer’s probability of purchase in various categories and expected market value or profit in future campaigns based on their purchase history and time-sequential pattern applicable to them. Third, based on the aggregation from these expected purchases, the methodology provides a quantitative customer market value that captures recency, frequency, and monetary measurements. Finally, an integer program model is developed that takes each customer’s expected market value in future campaigns as the input and optimizes the allocation of mailers to customers over multiple campaigns over the course of the year to maximize revenue. A decision support system based on the proposed methodology has been used for direct-mail marketing campaigns in Fred Meyer Jewelers since February, 2014. It has provided an effective and rationalized system and has significantly lifted customer redemption rate and increased jewelry revenues. A comparison of the same campaigns between 2013 and 2015 has shown an increase in revenue from $33M to $44M, or 33%. This increase has been achieved while the number of direct mailers has dropped from 5.6 million in 2013 to 2.8 million in 2015. The annual redemption rate has nearly tripled from 1.95% to 5.40%. The methodology has been a significant contributor to the increase in baseline performance of Fred Meyer Jewelers and could be applicable to other industries, such as consumer electronics, where long lead-times are present.
Page Count
136
Department or Program
Ph.D. in Engineering
Year Degree Awarded
2016
Copyright
Copyright 2016, all rights reserved. My ETD will be available under the "Fair Use" terms of copyright law.
