Determining Gender: A Database Marketing Imperative
“Men Are From Mars, Women are From Venus” by John Gray is the classic guide to understanding the opposite sex. This landmark book, first published in 1992, suggests that by learning how genders differ, men and women will be better able to communicate with and understand the opposite sex.
Over 25 years have passed since John Grey’s book became a bestseller and relationship reference guide, yet it continues to sell well as the differences between the sexes remain a mystery to many. Nowhere are these differences more apparent than in the ways that women and men purchase goods and services. For example, a long-held gender stereotype for new home seekers’ is supported by accounts from many real estate agents. They consistently report, “women are sold by a home’s kitchen, while men focus on the garage and other structural aspects during their first sales visit.” This knowledge has influenced sales offices and strategies for new home builders across the country.
Gender Marketing Research Studies
Differences between men and women’s shopping habits abound, providing market researchers with a consistent source of statistically different behaviors to cull from studies about retail shopping. “Men Buy, Women Shop,” is the title of a study undertaken by researchers at Wharton’s Jay H. Baker Retail Initiative and the Verde Group, a Toronto consulting firm. They found that many women love to shop, while men are more transactional by nature, focused on achieving their shopping objectives.
According to a separate study by First Insight,
a retail consulting firm, fewer men comparison price shop than women before purchasing and women are more frequently shopping on mobile devices than men. In addition, more men shop at full price retailers over discounters, consistent with the fact that fewer men are price comparison shoppers.
So, gender matters. While it is only one of several factors that make up the marketing genome (i.e., the list of attributes that help predict future buying behaviors), it is certainly one where we see significant differences in customer behaviors. Today’s leading marketers are making a concerted effort to collect as much relevant information that they can to better inform their marketing campaigns. Gender certainly falls into this category. On a database, then, how do we separate the men from the women?
Identifying Customer Gender: Reference Tables
There are many methodologies available to classify database records by gender. We’ll explore a few of them here.
The most common approach to determining a name’s gender is to use a reference table mapping first name to gender. For example, Mary is likely a female name and John is likely a male name. Some tables provide the relative probability that the name is male or female.
BuyerGenomics uses a “gender rank” to determine gender. Essentially, it determines the probability of a name being of a particular gender using values from a lookup table. If that probability is greater than 90%, then it\’s assigned that gender. If it\’s below 90%, it\’s ambiguous, and if the name does not appear in the table, it is unknown. Using tables alone, how effective can we be? Generally speaking, gender identification will classify approximately 80% of the names found on databases, leaving 20% of the names as “unknown” or “ambiguous.”
Social Security Administration: 138 years of gender data
As a quick aside, the social security administration has been cataloging the gender of names for social security applicants since 1880, so we have a comprehensive history of first names and gender as captured by the United States Government. These tables are one resource that has been used by compilers and allow us to assign relative probabilities of names to genders.The good news is that we have this history, but the bad news is that some names are simply ambiguous from a classification perspective. For example, Blake, Chase, Rory and many more can be considered unisex names. These names are normally classified as ambiguous, based on the percentage of names that are identified as male or female.
Parsing Honorific Titles for Gender Classification
In the absence of a classification, some data compilers and mailing service bureaus will use a customary honorific title, such as Miss, Mr., Ms., etc. to inform gender classification. The compilation of honorifics from surveys, warranty cards and other types of customer registrations has long been used to enhance classifications.
Prior to 1994, compilation of drivers’ licenses were also used for gender classifications. The Drivers Privacy Protection Act (DPPA) and subsequent amendments ended the use of this resource for marketers. Can you think of a better source for information about birth dates, corrective lenses, and gender? Probably not. Privacy laws will continue to impact publicly available data, a topic we will address in a future post.
Machine Learning and Gender Classifications
Finally, the advent of machine learning is a relatively new capability that is being used for gender classification. Known gender behaviors provide statistically significant predictors for the population of ambiguous gender database records. By examining overall behaviors, clicks, opens, and products purchased, gender classifications are being enhanced by big data applications.
The Marketing Genome, Gender and Marketing Campaigns
Mapping the human genome and understanding the implications about health and wellness is revolutionizing the way medicine is practiced. Similarly, addressable marketing, marketing databases, and big data applications are changing the way we think about marketing campaigns. In short, data science is allowing major marketers to map the marketing genome for their customers.
Understanding how gender impacts an individual\’s responsiveness to a marketing campaign is just one element that informs marketing campaigns and what we are referring to here as the marketing genome. Shopping behaviors such as comparison shopping and mobile purchasing are key elements to understand in managing the customer journey, and the correlation to gender was discussed above. There are thousands of additional variables that can inform marketing campaigns. Determining which variables to add and utilize vary by client and the type of customer relationship we are trying manage.
Customer Relationship Management
In addition to understanding gender behavior differences, the nature of the customer relationship is central to the success of any marketing campaign. In designing marketing campaigns, we need to answer several questions:
Answers to these questions and others provide the basis for creating an actionable customer intelligence platform.
- Who is the customer? Gender? What are the customer’s wants and needs?
- What is their geo-demographic profile? We may find many different segments here.
- From a lifestyle perspective, are there marketing opportunities?
- If they are a current customer, what have they purchased from us in the past?
- If they are a prospect, what is the most likely marketing approach to create trial of our good and services?
- Where is the customer vis-à-vis their customer journey? A buyer’s lifecycle can be pivotal in guiding 1-to-1 marketing campaigns
- How has a customer responded in the past to our marketing offerings?
Customer Intelligence Applications
At BuyerGenomics, we automatically assign a gender to each customer record, based on the ranking methodology described above. We combine this information with purchasing behaviors, psychographic and demographic data, and customer cohort information to guide marketing campaigns. BuyerGenomics allows marketers to:
Today’s customer intelligence and campaign management platforms are allowing marketers to capitalize on vast amounts of data and make informed decisions to guide marketing campaigns. As cloud computing has reduced the cost of new capabilities, marketers are more knowledgeable and empowered than ever before to manage customer relationships. Gender is a key element, but it is just the beginning.
- Identify who the most valuable customers are
- Understand the profile of customers
- Set up marketing experiments to determine the best marketing strategies by gender, marketing segment, or purchase patterns
- Prospect for new customers
- Reactivate dormant customers
- Create automated campaigns that respond to customer changes in behavior
Gary’s background includes over 30 years of analytics & database innovation for several leading Fortune 500 companies and Madison Avenue advertising agencies. Gary has been a frequent lecturer and author on the topics of database marketing and applied statistics. His articles have been published in DM News, Direct Marketing and the Journal of Direct Marketing. He recently was President of the Direct Marketing Idea Exchange and currently serves on their Board. Gary received his M.S. in Industrial Administration from Carnegie Mellon University.