CHAPTER 2

Developing Actionable Segmentation

Chapter Overview

Segmentation is widely considered the first step and key to understanding customers. However, in the real business world, the application of segmentation is not so effective. Many U.S. companies don’t use the segmentation scheme to guide decision making in acquisition, retention, resource allocation, and customer communications. This chapter first briefly discusses the importance and benefits of segmentation then explores why so many segmentations fail. Next it takes a deep dive into different levels of segmentation, and finally, it introduces several practical ways to develop meaningful tactical-level, managerial-level, and strategic-level segmentation.

This chapter is organized as follows:

• What Is Segmentation?
• Why Segmentation Failed?
• Three Levels of Segmentation
• How to Develop Tactical-Level Segmentation
• How to Develop Managerial-Level Segmentation
• How to Develop Strategic-Level Segmentation
• Conclusion

What Is Segmentation?

Segmentation divides customers into distinct, meaningful, and homogeneous subgroups based on one or multiple attributes and characteristics. The concept of segmentation has been the marketing mantra for many decades. Segmentation is widely accepted as a tactic that allows organizations to improve the effectiveness of their strategies and become more efficient with their expenditures. Benefits of segmentation include:

• Better Focus. All markets are heterogeneous; only a handful of companies can afford to serve the entire market; therefore, the majority of companies must identify and focus on segments where their products and services will have the highest chance of success.
• Higher Sales Revenue. Segmentation identifies the most lucrative opportunities. By spending more time and money on their most successful segments, companies can increase their win rate and ultimately increase revenues.
• Reduction of Marketing Cost. Segmentation helps optimize resources and allows companies to avoid spending their money and time on less lucrative opportunities and less profitable audiences.
• Better Product Development. Segmentation provides a clear idea of who wanted to buy a company’s product(s) and what features and functions they were looking for, which helps design and develop new products/services tailored to each segment’s characteristics rather than the mass market. Better product development will increase customer satisfaction and sales performance against competitors.
• Discovery of New Market. Segmentation verifies current targeting and positioning strategies. It reveals hidden problems of the current strategy and allows companies to identify underserved markets, so that they can fix issues in their strategy and enter the underserved markets.
• Personalized Communication. Segmentation helps marketers better understand the consumer differences in attitudes, needs, and behaviors. Equipped with these insights, marketers can structure different offers and messages accordingly to communicate with different consumers. These personalized communications will yield better brand impact than a uniform approach to communication across all customers.

The benefits of segmentation are great, and the concept of segmentation is easy to understand. However, segmentation is not so great in application. A study done by Harvard Business claimed that in the United States, 85 percent of 30,000 new product launches failed because of poor market segmentation.1

Why Segmentation Failed?

People generally blame the following issues:

1. The objectives of segmentation were not clearly identified before the segmentation study;
2. The wrong data or the inadequate data were used to develop segmentation;
3. The improper analytical techniques were used to produce the segmentation scheme;
4. Sales teams could not relate the personas in the segments to real customers;
5. The project didn’t get the support and sponsorship from top management;
6. Appropriate success metrics to measure the effectiveness of segmentation weren’t established;
7. Segmentation was too simple or too complicated for marketers to understand;
8. Segments were very hard to utilize to target specific customers for database marketing efforts. For instance, segmentation developed by psychologic and attitudinal analysis rather than a behavioral approach was difficult for marketers to link the segments to customers in the database.

While all these issues will certainly result in failed segmentation, in reality, most likely, the failures of segmentation initiatives generally are not related to the executive sponsorship because most of the segmentation projects were initiated by top management and were well funded. Inadequate data or incorrectly applied analytical techniques did cause the failures of segmentation, but this can be avoided by simply collecting more clean and relevant data and using alternative analytical techniques. Instead, the failures of segmentation were mostly because many marketers do not quite understand the levels of segmentation and tend to misuse high-level segmentation (i.e., market segmentation) to accomplish tasks that are supposed to be done by the lower level segmentation and vice versa. One common mistake is that the so-called one segmentation scheme fits all strategies. Some companies intentionally or unintentionally try to develop only one segmentation scheme and use it as a panacea to drive all marketing decisions. The problem is that different teams within the same company are faced with different challenges and have different needs. For instance, the chief executive officer (CEO) and the chief marketing officer (CMO) wanted to identify in what segment(s) the company has(have) the highest chance of success for their products and services, who these targeted customers are, where they live, and how to acquire them. Customer relationship management (CRM) directors were tasked with allocating the marketing budget to optimize customer management along a customer’s lifestages. Direct mail and e-mail campaign managers wanted to leverage segmentation data to personalize communications to each segment accordingly to increase response rates and sales and get the highest return on their marketing investments. And product managers wanted to achieve product-based goals, such as developing new products and selling more products. There is no such thing as one segmentation scheme that can solve every problem. Therefore, a company must build different segmentation schemes to address different needs and challenges facing various parties within the same organization.

I wanted to bring this critical issue to your attention for two reasons: First, misusing different levels of segmentation will result in lower marketing performance. And second, so far, only a few works of literature have discussed the concept of the levels of segmentation, which warranties a deeper dive into this topic.

So what exactly are different levels of segmentation?

Three Levels of Segmentation

Based on the business applications that segmentation schemes will influence, support, or drive, segmentation can be categorized into three levels:

The Strategic-Level Segmentation (Market Segmentation)

Strategic-level segmentation is also called enterprise-level segmentation or market segmentation, which has the goal of identifying market segments that differ in their demographics, purchasing power, goals, aspirations, and behaviors. Typically, a strategic segmentation view aims at understanding who the customer is, validating current positioning strategies, identifying products and market opportunities, and realignment of stores and employee behaviors around segments. Strategic-level segmentation links to the whole market, vision, strategic intent, and product benefits.

Managerial-Level Segmentation

Managerial-level segmentation takes multiple tactical-level segments as input, applying statistical methodologies such as recency, frequency, and monetary (RFM) analysis, lifetime value, and regression models to develop segments that help resources allocation, alignment, and planning.

Tactical-Level Segmentation

As the name implies, the purpose of tactical-level segmentation is used to complete tactical-level marketing tasks such as improving predictive analytics, increasing the response rate of direct mail campaigns, personalizing messaging, offers, and channels, and so on. Often, tactical-level segmentation is developed at a very granular level where customers in the database are sliced and diced in a number of ways: that is, by demographics (age, gender, income, ethnicity, etc.); by geography (country, state, city, zip, etc.); by the product affinity (product life stage, price, brand name, design, quality, features, functions, etc.); by behavioral data (RFM, seasonality, occasions, price proneness, communication channels, distribution channels, customer lifecycle stage, etc.); and by psychographic (consumer attitudes, motivations, values, behaviors, emotions, perceptions, beliefs, interests, etc.). Tactical segmentation can be simply developed by using just one variable or by using many variables combined. You may even include both strategic- and managerial-level segmentation data as input variables when developing multidimensional tactical segments.

Customer Segmentation Versus Market Segmentation

Tactical-level and managerial-level segmentation belong to customer segmentation. The differences between the customer segmentation and the market segmentation are:

1. Customer segmentation is developed by using customer data a company already holds. While data being used to create market segmentation are very broad and can include external data obtained from first and secondary research, and so on, market-segment research also typically includes prospects as well.
2. Market segmentation is typically used for high-level strategy, whereas customer segmentation offers a more detailed view for managing marketing budget and personalizing marketing programs.

How to Develop Tactical-Level Segmentation?

The journey of segmentation starts with building the tactical-level segmentation. Customers can be divided into the following six types of segmentation:

Behavioral Segmentation

Behavioral segmentation is a type of customer segmentation developed based on patterns of behavior displayed by customers as they interact with a company/brand or make a purchasing decision. It is about understanding customers not just by who they are, but by what they do. Behavioral segmentation allows businesses to divide customers into groups according to their knowledge of, use of, or response to a product, service, or brand.

The behavioral segments are the most important tactical-level segments because for some people, what they say might differ from what they will do. The best thing about behavioral segments is that they clearly reveal a customer’s true actions.

Behavioral segments, when used alone or together with other levels of segments, are instrumental in crafting personalized communications. Typical behavioral segments are as follows:

• Benefits Sought. What benefits of your products and services appeal to your customers? This information is tremendously helpful in product development and determines the right content for communications with prospects and customers.
• Source of Acquisition. This variable tells you what channel has acquired most customers and which channel has contributed the most sales revenue. For instance, the web channel may have acquired the most customers; however, you may also find that the average order value of customers acquired from the web channel might be lower than that of the brick and mortar store customers, and probably the margin is lower as well because many web customers were driven by promotions and deals. Marketers can use this piece of information to allocate the right acquisition dollars to each channel in a holistic and balanced manner.
• Distribution Channels. Grouping customers based on where they will go to purchase your product, online, store, or through a catalog. Multichannel customers normally spend more than single-channel customers. Distribution channel segmentation helps marketers to use the right channel to contact customers and develop retention programs to grow single-channel customers to multichannel customers and retain those most valuable multichannel customers.
• Seasonality. Seasons are opportunities for businesses to thrive. Each season brings unique holidays, events, and activities, which cause consumers to behave differently. Knowing customers’ buying patterns in different seasons helps marketers to send communications at the right time and provide customers the right products and offers.
• RFM. RFM is hands down the most important behavioral variable in data-driven marketing. They are not only the most critical differentiators in both tactical and managerial segmentations, but also essential components in any time-series models, logistic-response models, and customer lifetime value models.
• Margin and Profitability. Not all high monetary customers are profitable customers. For instance, some deals-only customers or bargain hunters could buy a lot. From a monetary value perspective, they may rank pretty high. However, their margin could be paper thin, or even negative, meaning the company didn’t make any money from them. The margin segments help identify the high- and low-margin customers so that the company treat them differently.
• Price Proneness. Marketers use price proneness data to personalize offers for customers. For instance, businesses don’t want to send full-price offers to the bargain hunters.
• Payment Tender Type. In general, credit customers will spend more than customers using other tender types.
• Affordability. If a company issues a proprietary credit card to its customers, knowing the open-to-buy amount will help predict how likely a customer will respond and how much more they can spend with this tender type. The open-to-buy amount is the difference between the credit limit assigned to a cardholder account and the present balance on the account.
• Tenure (Customer Loyalty). Tenure is defined as the days/months between a customer’s latest purchase date and his/her first purchase date. The tenure is usually used along with RFM. For instance, if the tenure of a high-value customer is 48 months but the recency of that customer is
  24 months, it simply means that the customer was loyal for four years before she/he became inactive in the past two years. A good practice is to send the customer several strong offers to reactivate her/him.
• Customer CRM Stage. This piece of information is critical to develop appropriate CRM contact strategies to onboard and grow new customers, retain and reactivate the best customers.
• Likelihood to Purchase. How likely a customer is going to make a buy in the future. This segment is usually developed based on response model scores, that is, the decile of model scores.
• Likelihood to Churn. How likely a customer is going to defect in the future. This segment is usually developed based on survival model scores.
• Predicted 12-Month Purchase Amount. Knowing how much a customer will spend in the next 12 months is critical in determining the right contact frequencies for each managerial segment to maximize response and sales revenue of this customer.

Geographic Segmentation

Geographic segmentation groups customers by a specific area, such as regions of the country or state and urban or rural. Based on specific business needs, a business may consider developing the following geographic segments: climate zone, region, state, district, rural area, city, and zip groups.

• Region. Region identifies where the customer lives, that is, East, West, mid-West, and so on.
• State/Province. That is, California, Texas, Ontario, and
  so on.
• Climate Zone. The U.S. climate zones are typically grouped into five different regions: the Northeast, the Southwest, the West, the Southeast, and the Midwest. U.S. climate varies dramatically by region. Stores in different climate zones need to adjust their inventory assortments to meet local customer needs. Also, the weather allows marketers to connect and communicate with consumers on a highly local level, targeting them with a relevant message at the right time. For instance, when inclement weather is approaching, marketers can highlight a product’s superior performance in such conditions.

Demographic Segmentation

Demographic segmentation groups customers by age, generation, gender, income level, occupation education, marital status, religion, race, family size, and so on. To use demographic data in segmentation work, a business will have to buy them from a data vendor such as Claritas Prizm, Experian, Acxiom, or even census data. When limited to purchasing only a few demographic variables due to budget constraints, I recommend at least buying the top three variables: age, gender, and household income.

• Age. Age is one of the top three most important demographic variables, in my opinion. Narrowing down which age group(s) contributes(contribute) most to sales and what products and features appeal to them will significantly help improve the efficiency and effectiveness of marketing communications. How to break out the age groups depends on the ages of customers. Also, one thing I’d like to remind analysts is that if your target customers are mainly teenagers, then the age information captured in the database might not be so accurate. Because often it was the parents who paid the bill for their children; therefore, the ages obtained in the database might be parents’ ages not the end-users’ ages.
• Generation. Generation helps marketers understand how different generations like to consume information and what features appeal to them the most.
• Gender. Like age, gender is another critical piece of information. Even with the same segment, male and female customers will almost always display different purchase behaviors.
• Household Income. Income is highly associated with affordability. The value and the prices of products and services determine the income levels of customers. When reviewing income-level data, two things to pay special attention to are: (1) Some low-income-level customers might be the top spenders. A closer look at these customers at a household level will help to better understand their home locations, social status, and lifestyles. (2) If the primary target customers are middle-level Americans with an income level ranging from $25,000 to $75,000, but somehow a significant number of best customers have a household income higher than $100,000—does that mean a business needs to pursue more affluent customers? Not necessarily. It may have to look at their discretionary buying power as well. A family living in San Jose, California with an annual income of $125,000 might have the same discretionary buying power as one living in Waco, Texas that makes only $60,000 a year.
• Marital Status. It is quite understandable that the needs of a married man/woman are very different from that of a single man/woman.
• Occupation. Certain professions may be particularly fond of certain products or services. For instance, many sales professionals like to play golf with their clients. They might be the best customers for sports retailers.
• Education. People with different education levels display different content preferences and buying power, which helps marketers determine the right offers, advertising channels, and contents to communicate with their customers effectively.
• Ethnicity. Customers of some races will show favoritism to certain products. For instance, jewelers know that Indian American customers and Hispanic customers like gold jewelry. Therefore, if a business has stores near these customers, it may want to adjust the merchandising assortments in these stores to better satisfy the needs of these ethnic groups.

Psychographic Segmentation

Psychographic segmentation groups customers into cultural clusters, social status, and lifestyle.

• Prizm Life Stages. Claritas PRIZM Premier is a set of geodemographic segments for the United States, developed by Claritas Inc. PRIZM Premier classifies every U.S. household into one of 68 consumer segments based on the household’s purchasing preferences. According to Claritas,

PRIZM Premier Lifestage Groups account for affluence and a combination of householder age and household composition. Within three Lifestage classes—Younger Years, Family Life, and Mature Years—the 68 segments are further grouped into 11 Lifestage Groups. Each Lifestage Group’s combination of the three variables—affluence, householder age, and presence of children at home together offer a more robust picture of the consumer.2

• Claritas PRIZM Premier Social Groups. These 14 social groups are based on Claritas’ urbanization class and affluence, two important and unique variables used in the creation of PRIZM Premier. There are four urbanization class categories—urban, suburban, second city, or town and rural. Within each of these categories, all the segments are then sorted into groups based on affluence, another powerful demographic tool when understanding consumer behavior and motivation. If the marketing budget allows, I will strongly encourage businesses to append the Prize Premier to customer data (disclaimer: I have no business relationship with Claritas Inc.), because they are very handy when developing customer profiles and buyer personas.

Product Segmentation

Product segmentation groups customers into product categories purchased. Please be advised that segments in this type of segmentation are not mutually exclusive. The same customer may appear in multiple segments. Product segments reflect a customer’s historical needs and help predict the next product/service the customer will purchase.

• Product Department Segmentation. This is the top level of the merchandise classification tree. Product department segments help identify what department(s) contributes(contribute) the most to the number of customers and to the sales revenue.
• Subdepartment. Under each product department category, a business may further drill down to see how each subcategory performs.
• Cross-buying Combinations. This segment helps pair more relevant product offers to meet the needs of customers/prospects, thus increasing cross-sell success rate and sales revenue.
• Product User Types. Separating user types enable marketers to provide more personalized and relevant messages, products, and services to meet customers’ needs better, improving both customer experience and sales revenue.
• Brand Name. Segmenting customers based on the brand name can help identify customers’ brand preferences.

How to Use Tactical-Level Segmentation

Tactical-level segmentation depicts customers from a great variety of perspectives. At this point, a business should be able to gain a pretty good idea about its customers already.

Applications of tactical-level segmentation include but are not limited to advertising, sales, and promotions (direct mail, catalog, e-mail, display, TV, etc.). Tactical-level segments can be used to:

• Help discover the key attributes of customers and provide insights into customers in terms of demographics, geographic, needs, and behaviors. Try cross-tabulating variables and test different combinations or use factor analysis and clustering analysis to develop more sophisticated tactical segments. A company will gain a much deeper understanding of its customers.
• Serve as input variables for improving enterprise-level segmentation work that was developed using traditional qualitative methods such as survey and focus groups. Insights gleaned from tactical segmentation can be used to compare to insights gained from qualitative analysis such as customer survey and focus group, improving the accuracy of strategic-level segmentation.
• Serve as key variables for predictive modeling. Each segment is a good descriptor. A business may use the original value of tactical segments or transform them into different variables. For instance, recency can be broken out into three categorical variables such as active, inactive, and dormant.
• Personalize communications. Segments can be used to narrow down and refine mail lists, determine the right offers, price points, and products.
• Facilitate trigger marketing communications. For instance, when an active customer was turning to inactive, a “We Miss You” e-mail with a stronger and personalized offer will be triggered to encourage the customer to buy again.

Tactical segments provide a customer snapshot. To derive deeper insights from tactical segments, a business needs to look at them in a holistic and longitudinal view. I like to run these tactical segmentation reports quarterly. Keeping track of them for 24 or 36+ months will surely catch many early signs of threats and discover many hidden growth opportunities as well.

How to Develop Managerial-Level Segmentation

The objectives of developing a managerial segmentation are: (1) to help allocate marketing dollars based on projected return on investments (ROIs), (2) to improve the effectiveness of CRM based on a customer’s life stage and expected future value, and (3) to serve as a marketing automation tool for triggered marketing communications when combined with tactical segmentation and/or predictive modeling.

Managerial Segmentation for Retention

There are many ways to build managerial-level segmentation for retention; one of the useful ways is Mu’s RFM model.

Mu’s RFM Model was inspired by Arthur Hughes’ 125 RFM segmentation model. Mr. Hughes’ 125 segments are a useful segmentation tool, but 125 segments are not easy to manage, so I came up with a new way to simplify them. My model starts with segmenting the database using three key variables: Recency, Frequency, and Monetary.

Recency: How recently did a customer make a purchase?

For instance, I like to divide the recency into three groups:

• R1: recency between 0 and 12 months (active customers);
• R2: recency between 13 and 24 months (inactive customers);
• R3: recency >24 months (dormant customers).

This breakup is purely for illustration purpose. Based on customers’ purchase cycle and business characteristics, different cut-off dates, such as, 3 months, 6 months, or even 24 months can be used to define active customers. Another option is to normalize them into percentiles. For instance, categorizing the most recent 20 percent customers as active, the next 30 percent customers as inactive, and the rest as dormant.

Frequency: How often does a customer make a visit or a purchase?

Frequency is divided into three groups, too:

• F1: customers who have paid three-plus qualified visits;
• F2: customers who have paid two qualified visits;
• F3: customers who have paid one qualified visit.

Isolating one-visit customers is very important as it enables marketers to develop specific programs to convert them into two-visit, a critical step to increase customer retention. A qualified visit is usually defined as one that has the minimum purchase amount meeting or exceeding a threshold, let’s say $99, for example.

Monetary: How much does a customer spend on a purchase(s)?

Monetary is also divided into three groups:

• M1: customers whose total spending is among the top 30 percentile;
• M2: customers whose total spending falls into the middle 40 percentile;
• M3: customers whose total spending is at the bottom 30 percentile (Table 2.1).

Table 2.1 Mu’s 27 RFM combinations

	Recency	X		Frequency	X		Monetary
R1	0-12 Months		F1	3+ Visits		M1	Top 30%
R2	13-24 Months		F2	2 Visits		M2	Middle 40%
R3	>24 Months		F3	1 Visit		M3	Bottom 30%

Now crosstab recency with frequency and monetary, there are 27 segments (3 × 3 × 3 = 27, see Table 2.2), where 111 stands for R1, F1, and M1. This is the group of customers with the most recent purchase, most visits, and most monetary value; and 333 stands for R3, F3, and M3, which consists of the dormant, one-time purchase, low monetary customers. From a marketing perspective, this is the group of customers a business doesn’t want to spend additional time and money on.

Table 2.2 Mu’s 27 RFM segments

	Life Stage
Frequency	R1: Active	R2: Inactive	R3: Dormant	Monetary
F1: 3+Visits	111	211	311	M1: High
	112	212	312	M2: Medium
	113	213	313	M3: Low
F2: 2 Visits	121	221	321	M1: High
	122	222	322	M2: Medium
	123	223	323	M3: Low
F3: 1 Visit	131	231	331	M1: High
	132	232	332	M2: Medium
	133	233	333	M3: Low

These 27 segments essentially are a list of lists that can be used as mail lists for direct mail, catalog, and e-mail campaigns. Normally segment 111 (R1, F1, and M1) will deliver the highest response rate and the highest sales revenue as well, whereas segment 333 (R3, F3, and M3) will generate the lowest response rate and the least sales revenue.

One of the major uses of managerial segmentation is to allocate the marketing budget. If the 27 segments are still too many to manage for
this matter, they can be further consolidated into six larger groups (see Table 2.3) based on projected response rate and sales revenue.

Table 2.3 Six RFM groups

Active	Inactive	Dormant	Monetary
A1: 111, 121, 131	I1: 211, 221, 231	311, 321, 331	High
A2: 112, 122, 132	I2: 212, 222, 232	312, 322, 332	Medium
A3: 113, 123, 133	I3: 213, 223, 233	313, 323, 333	Low

Now two actions may be taken to these six groups:

First, assigning customer management goals to each group (Table 2.4).

Table 2.4 Assigning goals to RFM groups

Segment Group	RFM Segment	Customer Management Goals
A1	111, 121, 131	Aggressively Retain and Grow
A2	112, 122, 132	Retain and Grow
A3	113, 123, 133	Selectively Grow
I1	211, 221, 231, 311, 321, 331	Reactivate and Grow
I2	212, 222, 232	Selectively Reactivate and Grow
I3	213, 223, 233, 312, 322, 332, 313, 323, 333	Natural Course

For instance, Group A1 is the most recent and also the highest monetary group. The CRM goal is to retain and further grow customers in this group aggressively. The total number of customers in this group may be less than 10 percent of total customers, but they often contribute more than 30 percent of total sales revenue. Retaining these customers as long as possible is critical to both the top line and the bottom line of a business.

Group I3 may contain a lot of low monetary inactive and dormant customers. Analysis of their contact history data may reveal that many of them had been contacted multiple times but never responded to communications. So, don’t waste limited marketing dollars on this group of customers. The strategy is to treat them as prospect lists and only contact them selectively when the budget allows.

CRM managers/directors can use this segmentation scheme to determine how many marketing dollars should be spent on each segment group based on the projected sales revenue of each segment group.

One of the drawbacks of this methodology is that sometimes, a high monetary customer may not necessarily guarantee a high future value. For instance, in the jewelry business, a male engagement ring customer after his first big-ticket purchase may not come back and purchase again for many years. That is a typical example of a high past value but low future value customer.

To overcome this issue, the second thing to do is overlay the projected future value onto the six groups created.

How to Calculate the Projected Future Value

First, determine the timeframe for forecasting future value. Calculate the predicted future value for the next 6 months, 12 months, or even 24 months. The timeframe of prediction should be determined based on the length and patterns of customers’ purchase circles.

In Chapter 8, we introduce several ways to calculate the future value for each existing customer. One of the easiest ways is the two-model approach. Basically, a business will need to build two models: (1) a logistic regression model that predicts the likelihood of purchase in the next 12 months and (2) a regression model that predicts the total spend in the next 12 months.

The final predicted value = the predicted likelihood to purchase × the predicted total spends in the next forecasted period.

The projected value can be divided into two groups: (1) the high-value group that consists of the top 30 percentile of customers and (2) the low-to-medium group that has the rest of the 70 percentile customers. Why is the database split this way? Remember the famous 80/20 rule—20 percent of customers produce roughly 80 percent of the sales? But in reality, it’s more likely the top 30 percent of customers will contribute a little bit more than 80 percent of sales. That is why we do a 30/70 split.

After applying the predicted future value to the RFM segment groups, the number of segments will increase, but it will enable a business to set more appropriate goals and develop more accurate contact strategies for each segment. For instance, segments high-A1 and low-to-medium-A1 before the split belonged to the same group. After the split, the high-A1 group is more valuable; therefore, it will demand more resources and attention from marketers.

Active Customer Segment Groups

Six active RFM groups broken out by projected customer lifetime value are shown in Table 2.5.

Table 2.5 Six active RFM groups broken out by projected customer lifetime value

Future Value	Segment Group	RFM Segment	Customer Management Goal
High	A1	111, 121, 131	Aggressively Retain and Grow
	A2	112, 122, 132	Aggressively Retain and Grow
	A3	113, 123, 133	Retain and Grow
Low2Medium	A1	111, 121, 131	Retain and Grow
	A2	112, 122, 132	Selectively Retain and Grow
	A3	113, 123, 133	Selectively Retain and Grow

Inactive Customer Segment Groups

Six inactive RFM groups broken out by projected customer lifetime value are shown in Table 2.6.

Table 2.6 Six inactive RFM groups broken out by projected customer lifetime value

Future Value	Segment Group	RFM Segment	Customer Management Goal
High	I1	211, 221, 231, 311, 321, 331	Aggressively Reactivate and Grow
	I2	212, 222, 232	Reactivate and Grow
	I3	213, 223, 233, 312, 322, 332, 313, 323, 333	Selectively Reactivate and Grow
Low2Medium	I1	211, 221, 231, 311, 321, 331	Natural Course
	I2	212, 222, 232	Natural Course
	I3	213, 223, 233, 312, 322, 332, 313, 323, 333	Natural Course

Predicted Future Value Model + Customer Life Stage Model

Another variation of this RFM-based segmentation scheme is to separate the new customers from the active customers and combine inactive and dormant customers into one group. I call this approach the Predicted Future Value Model + Customer Life Stage Model.

In this model, new customers are defined as recency ≤ 30 days (can also be set as ≤60 days or however the business usually defines them). We don’t further break them down by monetary because usually the number of new customers won’t be huge.

Active customers were divided into three segments by monetary—high, medium, and low.

Inactive and dormant customers were divided into two groups: high monetary and low-to-medium monetary. Singling out the high monetary customers from this group is extremely important as we only want to spend money to reactivate these customers.

After overlaying the projected future value onto these six groups, there are 12 segments (see chart below), which is still very manageable.

The New Customer Group has two segments:

• 1: New Customer High Future Value;
• 2: New Customer Low-to-Medium Future Value.

The Active Group has six segments:

• 3: Active High Monetary High Future Value;
• 4: Active High Monetary Low-to-Medium Future Value;
• 5: Active Medium Monetary High Future Value;
• 6: Active Medium Monetary Low-to-Medium Future Value;
• 7: Active Low Monetary High Future Value;
• 8: Active Low Monetary Low-to-Medium Future Value.

The Inactive and Dormant Group has four segments:

• 9: Inactive High Monetary High Future Value;
• 10: Inactive High Monetary Low-to-Medium Future Value;
• 11: Inactive Low Monetary High Future Value;
• 12: Inactive Low Monetary Low-to-Medium Future Value.

Next, you will determine the contact strategies for each segment (Table 2.7).

Table 2.7 Life stage + Predicted future value

Life Stage
Future ValueHistoric Value	New		Active		Inactive and Dormant
	All	High Monetary	Medium Monetary	Low Monetary	High Monetary	Low2­-Medium
High	1: New-HF	3: Active-HM-HF	5: Active-MM-HF	7: Active-LM-FH	9: Inactive-HM-HF	11: Inactive-L2MM -HF
Low2Medium	2: New-L2MF	4: Active-HM-L2MF	6: Active-MM-L2MF	8: Active-LM-L2MF	10: Inactive-HM-L2MF	12: Inactive-L2MM-L2MF

For example, High Monetary, High Future Value customers will receive the most aggressive contact frequencies. New High Future Value customers will receive aggressive onboarding communications (Table 2.8).

Table 2.8 Life stage + Predicted future value + Goals

Life Stage
Future ValueHistoric Value	New	Active			Inactive and Dormant
	All	High Monetary	Medium	Low	High	Low2-Medium
High	Aggressively Onboard	Aggressively Retain and Grow	Aggressively Retain and Grow	Retain and Grow	Reactivate	Selectively Reactivate
Low2Medium	Onboard	Retain and Grow	Retain and Grow	Natural Course	Selectively Reactivate	Natural Course

Applications of Managerial Segmentation

The managerial segments can help allocate limited marketing dollars and achieve higher ROIs in both customer acquisition and retention programs. First, let’s look at a few things managerial segments can do to reduce mail cost and increase the effectiveness of direct mail and catalog campaigns.

Application One: Determining the Right Mail Quantities for Direct Mail and Catalogs

A national retailer used to proudly send their catalogs to every customer in the database, a practice that any experienced marketers will surely frown upon. So, before they decided the mail quantity for the upcoming catalog this year, I pulled the matchback analysis report of last year’s catalog and overlaid the data on the top of managerial segments (Table 2.9). The campaign results are as follows (for illustrative purpose only, data have been altered).

Table 2.9 ABC company holiday catalog matchback analysis

Segment	Mailed	# Responders	RR	Cost/Pcs	Total Cost	AOV	Total Sales	Total Margin	ROI
3: Active-HM-HF	30,000	2,400	8.00%	$1.50	$45,000	$450	$10,80,000	$4,32,000	860%
5: Active-MM-HF	60,000	3,600	6.00%	$1.50	$90,000	$342	$12,31,200	$4,92,480	447%
1: New-HF	10,000	500	5.00%	$1.50	$15,000	$300	$1,50,000	$60,000	300%
4: Active-HM-L2MF	40,000	1,600	4.00%	$1.50	$60,000	$150	$2,40,000	$96,000	60%
2: New-L2MF	20,000	700	3.50%	$1.50	$30,000	$135	$94,500	$37,800	26%
7: Active-LM-HF	15,000	450	3.00%	$1.50	$22,500	$280	$1,26,000	$50,400	124%
9: Inactive-HM-HF	20,000	600	3.00%	$1.50	$30,000	$358	$2,14,800	$85,920	186%
6: Active-MM-L2MF	70,000	1,750	2.50%	$1.50	$1,05,000	$130	$2,27,500	$91,000	-13%
10: Inactive-HM-L2MF	15,000	225	1.50%	$1.50	$22,500	$220	$49,500	$19,800	-12%
8: Active-LM-L2MF	50,000	500	1.00%	$1.50	$75,000	$80	$40,000	$16,000	-79%
11: Inactive-L2MM-HF	12,000	60	0.50%	$1.50	$18,000	$330	$19,800	$7,920	-56%
12: Inactive-L2MM-L2MF	90,000	180	0.30%	$1.50	$1,35,000	$78	$14,040	$5,616	-96%
Total	4,32,000	12,565	2.9%	$1.50	$6,48,000	$278	$34,87,340	$13,94,936	115%

Overall, the catalog generated $1,394,936 margin and a 115 percent ROI. However, if you remove the bottom three low response segments (highlighted in yellow), the ROI significantly improves from 115 percent to 225 percent (see Table 2.10).

Table 2.10 ABC company holiday catalog matchback analysis—high and medium response groups

Segment	Mailed	# Responders	Response Rate	Cost/Pcs	Total Cost	Avg Sales	Total Sales	Total Margin	ROI
3: Active-HM-HF	30,000	2,400	8.00%	$1.50	$45,000	$450	$10,80,000	$4,32,000	860%
5: Active-MM-HF	60,000	3,600	6.00%	$1.50	$90,000	$342	$12,31,200	$4,92,480	447%
1: New-HF	10,000	500	5.00%	$1.50	$15,000	$300	$1,50,000	$60,000	300%
4: Active-HM-L2MF	40,000	1,600	4.00%	$1.50	$60,000	$150	$2,40,000	$96,000	60%
2: New-L2MF	20,000	700	3.50%	$1.50	$30,000	$135	$94,500	$37,800	26%
7: Active-LM-HF	15,000	450	3.00%	$1.50	$22,500	$280	$1,26,000	$50,400	124%
9: Inactive-HM-HF	20,000	600	3.00%	$1.50	$30,000	$358	$2,14,800	$85,920	186%
6: Active-MM-L2MF	70,000	1,750	2.50%	$1.50	$1,05,000	$130	$2,27,500	$91,000	-13%
10: Inactive-HM-L2MF	15,000	225	1.50%	$1.50	$22,500	$220	$49,500	$19,800	-12%
Total	2,80,000	11,825	4.2%	$1.50	$4,20,000	$289	$34,13,500	$13,65,400	225%

This analysis convinced the CMO of the company to reduce the mail quantity by 35 percent in 2017 and invest the money saved from the catalog mailing into other more productive marketing channels such as display and remarketing. That small adjustment in budget allocation alone has generated more than $200K incremental sales for the company in 2017.

Application Two: Setting Up Contact Limits to Reduce Offer Fatigue

Many retailers have over mailed their customers, resulting in a huge waste of money and offer fatigue as well. I have seen some retailers send out as many as 20+ direct mail and catalogs and almost 50+ e-mails every year. Therefore, some retailers decided to cap the maximum number of contacts per month a customer can get. For instance, one can only get two direct mails a month and only get the same offer every three months. That is a good practice; however, customers are not created equal. A universal contact number for all customers might not be the best solution to maximize the ROIs of marketing programs. This is when managerial segments kick in and help.

Table 2.11 indicates that the number of contacts each segment will receive is different. For instance, the High Monetary and High Future Value segment will get eight contacts, the most number of contacts of all the groups, whereas three inactive and low future value groups (segments 8, 10, and 12) will only get e-mail communications.

Table 2.11 Contact strategy for RFM groups

Life Stage
Contact Policy	New	Active			Inactive and Dormant
Future ValueHistoric Value	All	High Monetary	Medium Monetary	Low Monetary	High Monetary	Low2Medium
High	1: New-HF: Max 6 Contacts; 2 Same Offers / Month, Block all other offers for the first 4 weeks, etc.	3: Active-HM-HF 8 Contacts/year; 2 Same Offers/Month, Relevant Offers Only	5: Active-MM-HF; 6 Contacts/year, 2 Same Offers/Month, Relevant Offers Only	7: Active-LM-FH; 3 Contacts/year, Relevant Offers Only	9: Inactive-HM-HF; 4 Contacts/year, Relevant Offers Only	11: Inactive-L2MM-HF Use Response Model for mail customer selection; 3 Contacts/year, Relevant Offers Only, 15 Miles Shopping Range
Low2Medium	2: New-L2MF; 3 Contacts/Year, Block all other offers for the first 4 weeks, etc.	4: Active-HM-L2MF; 6 Contacts/year, 2 Same Offers/Month, Relevant Offers Only	6: Active-MM-L2MF Use Response Model for mail customer selection; 4 Contacts/year, 2 Same Offers/Month, Relevant Offers Only, 15 Miles Shopping Range	8: Active-LM-L2MF Email Offers Only	10: Inactive-HM-L2MF Email Offers Only	12: Inactive-L2MM-L2MF Email Offers Only

Application 3: Developing Best Customer Profiles for New Customer Acquisition

Customer acquisition is a driver for businesses. But customers are not created equal. To reduce acquisition cost and increase the effectiveness of acquisition efforts, a business must first know what customers will generate the highest customer lifetime value. Overlaying customer lifetime value data on the top of the managerial segment will give marketers deeper insight into what kind of prospects they want to acquire. A business may want to acquire prospects that look like the following three types of customers because their lifetime values are the highest among all segments:

• High Historical Monetary and High Future Value. These are the best customers to acquire;
• High Historical Monetary but Low Future Value. These customers are critical to the sales revenue of near terms, that is, the next year;
• Low-to-Medium Historical Monetary but High Future Value. This group of customers didn’t spend as much as the High Historical Monetary but Low Future Value (HM-LF) customers, but long term, they are likely to buy a lot. They are not as good as the other two types of customers but are undoubtedly worth acquiring if the marketing budget allows.

So basically, a business should develop customer profiles for these three segments (there might be some overlaps among these three segments, though) and then find prospects in the target markets that look like them.

Other Marketing Applications

Using both tactical segments and managerial segments separately or combined, marketers can develop a variety of triggered marketing automation programs and contact cadences based on customer management goals. For instance, they may set up onboarding triggers, cross-sell and up-sell triggers, repurchase triggers, lapse triggers, anniversary triggers, and many more for e-mail and direct mail campaigns.

How to Develop Strategic-Level Segmentation

When businesses develop their marketing strategies and decide how to create value for customers, there are essentially two issues they must consider: (1) they must determine which segment(s) they have the best chance of commercial success to sell their products and services to and (2) they must choose the right product-positioning strategies to compete in the segment(s) they have chosen and develop individualized products/services tailored to target segments and finally decide on the optimal marketing mix for each selected segment.

Marketers solve these two issues by using a model called STP,3 which stands for segmentation, targeting, and positioning. Segmentation is the first element and also the first step of the STP modeling.

The key outcome of the segmentation, targeting, and positioning process is to develop a unique marketing mix for a specified target market. If a segment is not responsive to a distinct offering, then that segment can probably be combined with another similar segment.

What Is Market Segmentation?

Most literature, when talking about segmentation, refers to the market segmentation. The market segmentation is an enterprise-level segmentation or strategic-level segmentation, which has the goal of identifying market segments that differ in their demographics, purchasing power, goals, aspirations, and behaviors. Typically, a strategic segmentation view aims at understanding who the customer is, validating current positioning strategies, identifying products and market opportunities, and realignment of stores and employee behaviors around segments. Market segmentation links to the whole market, vision, strategic intent, and product benefits.

Criteria for Market Segmentation

A good market segmentation must meet the following four criteria:

• Homogeneous within the group and heterogeneous between the groups. The number one rule of segmentation is to make sure that customers within each segment are similar in terms of needs or other characteristics such as income level, interests, buying behavior, location, and so on. Differences between each segment of consumers should be distinguished and significant.
• Sizable. The market segment should be large enough in terms of the number of potential customers, sales revenue, and profitability to ensure the company can achieve sufficient financial return.
• Measurable. The key performance indicators should be available to measure the financial success of the market segment.
• Actionable. Segments must enable businesses to implement a distinctive marketing mix for each market segment. Each market segment should respond better to a distinct marketing mix, rather than a generic offer.

Why Market Segmentation Failed

Earlier in this chapter, we listed eight common reasons that lead to failures of segmentation. In addition to these reasons, there are a few other causes worth mentioning.

One of the commonly seen issues of developing strategic segmentation is that the segmentation study was heavily influenced by the opinions and the objectives of top executives. Sometimes, the purpose of the segmentation study was simply to fortify the executive’s planned strategies. A typical example was in January 2012, just two months after Ron Johnson took over the helm of the 110-year-old department store JCPenney, he announced in an interview that “We want to be the favorite store for everyone, for all Americans, rich and poor, young and old,”4 indicating a significant shift of the company’s targeting strategy. While there are a few companies that can afford to serve the entire market, no department store has ever succeeded by targeting all customers. Surprisingly, the media were quiet about his strategy. Only Alexander Chernev, an associate professor of marketing at the Kellogg School of Management, Northwestern University, questioned JCP’s desire to please everybody. Mr. Chernev says it goes against conventional wisdom.5

A few years back, I worked for an international specialty retailer that hired a big consulting firm to develop a new strategic segmentation. The consulting firm surveyed 2,000 plus people and developed a segmentation scheme without looking at the internal customer data. Having read through the beautiful PowerPoint presentation deck put together by the consulting firm, I found that the segments they created just could not be validated by the customers in the database. Many marketing managers within the company didn’t believe the personas either because they did not look like the customers they were familiar with. Later I understood that the CEO of the company planned to change the current marketing strategy. He wanted to pursue customers who look like the personas described in the segmentation work. So, the segmentation study was more of a tool to help achieve his political agenda. I wondered if the consulting firm had checked the real customer data first whether they would have developed a different segmentation scheme.

How to Develop Market Segmentation

Unless they have a strong internal marketing research team, considering the importance and sophistication of building a strategic-level segmentation, companies usually outsource market segmentation studies to a consulting firm. However, that does not necessarily mean the marketing and analytics team don’t need to do anything and leave everything to the service provider. The success of a market segmentation project largely depends on active participation. The company can help the service provider do a better job by:

1. Clearly Outlining the Objectives of Segmentation Study

   For most established businesses, chances are they have already developed a market segmentation scheme. Due to the fast-changing retail environment and the proliferation of customer data that may provide more insights into customers and the markets, it is highly recommended that companies review their strategic segmentation once every three to five years.

   Determining the objectives of segmentation starts by auditing the current segmentation scheme. A marketing strategy questionnaire is a great tool to ask the management team to:

   • describe the markets and existing customer segments the company is competing in. What market segments are you targeting (list segment name and characteristics)? What are its members’ demographics and psychographics? What is the current size of the segment in terms of population and dollar amount? What is the customer’s primary reason for buying or wanting to use our product or service? Are there any issues or concerns that the target audience might have regarding the products or services?
   • describe the company pricing strategy. How important is price in the decision process?
   • describe marketing communication strategies;
   • describe strength, weakness, threats, and opportunities in each segment;
   • identify what industry trends can inhibit success;
   • describe key business needs of the potential segments and the sizes of these segments in terms of population and dollar amount.

Through this exercise, marketers will be able to gain the following benefits:

• Better understanding customers’ needs and wants, their feelings and thinking, their behaviors, and their future move;
• Determining whether products and services can satisfy their needs in the selected segments;
• Determining whether marketers have effectively communicated the value of the products and services to the selected segments;
• Confirming whether the current targeting and positioning strategies are working;
• Identifying new market(s) that is(are) financially attractive;
• Uncovering hidden opportunities and threats;
• Gaining better ideas to realign resources, stores, and employee behaviors around targeted segments;
• Developing a look-like customer model for acquisition based on segment profiles and customer personas;
• Identifying the gap and issues in current information systems;
• Developing a hypothesis. During the internal audit phase, some managers will also develop a hypothesis about the ups and downs of the business. A hypothesis is supposition, explanation, or theory to interpret certain events or phenomena, yet needs to be verified or corroborated by the segmentation study. In marketing, the hypothesis is often centered on changes in target customers, product, price, location, channel, competition, communications, macroeconomy, and so on. For instance, a business may speculate that the size of a specific segment ABC will grow faster than the size of the segment DFG; it believes that segment ABC will be the next growth opportunity for the company. Or the business speculates that the reason why the new home sale has increased by 30 percent in the past two years was that more millennials started buying homes, so they need to spend more marketing dollars targeting millennials. Or the diamond jewelry sales decreased because younger consumers spent more money on other precious stones; therefore, they need to adjust the product assortments, and so on.

Equipped with insights gained through the auditing of the current segmentation scheme, the company will have a much better idea about what it wants to get out of the market segmentation study. Having the objectives clearly outlined is the first step and key to the success of the market segmentation project.

2. Ensuring the Research Data Are Enough and of High Quality

   Research firms use interview and survey as two major means to gather data and form psychographic profiles for people interviewed or surveyed. In-depth interviews allow for collection of useful qualitative data to understand what makes customers tick. But it is costly. A survey may not be able to get as insightful answers as interviews, but it reaches more people than interviews. The online survey has quickly gained popularity recently as it is quicker, cheaper, and more convenient than the traditional telephone and mail-in survey. However, marketers need to be careful using this vehicle. For instance, if the primary customers are senior citizens, an online survey might not be an appropriate way to collect data as the results will skew toward younger customers. The reason is that many senior citizens are not very familiar with the online survey. A real case was eight years ago. One well-known golf equipment manufacturer presented its newest and coolest golf driver to us. They recommended targeting the 28- to 45-year-old male group. Based on customer data, we knew that their customers were typically 40- to 65-year-old male. So, I asked them why they shifted their focus toward younger customers. They said they drew the conclusions based on the finding of a recent online survey. A deeper dive into the population of those who took the online survey found that they were mostly younger male golfers because they accounted for 70 percent of people who took the online survey. Therefore, the demographic skews toward younger golfers. By the same token, if the majority of customers are youth, then Facebook might not be an appropriate place to survey them. The lesson I’ve learned from that case is to work closely with the research company and help them truly understand the nature of the business and the characteristics of customers, which will help them to choose the right channel and vehicles to collect data.

   The second thing worth mentioning is that while these professional research service companies can help avoid most survey design errors, special attention must be paid to the following commonly seen survey errors:

   • Does the questionnaire assume prior knowledge or understanding? The survey should not rely on presumed prior knowledge from respondents. Try to avoid utilizing acronyms or industry jargon that may not be readily known by all customers.
   • Does the questionnaire include leading questions? Leading questions are ones that supply the facts or suggest the answer in the wording of the questions; the question itself can “lead” respondents to a particular response. For example, “Company XYZ has teamed up with the world-famous fashion designer Mr. X to design the world-class brand name dresses for women. What are your thoughts on these first-class products?”
   • Does the survey have too many open-end questions? While open-end questions often provide the most useful insights, dealing with hundreds or thousands of answers, and coding the method of turning qualitative data into insights could be challenging.
   • Do they survey enough people so that the outcomes are statistically significant? If you haven’t surveyed enough people, you will not be able to derive meaningful insights from the data collected.
   • How they handle survey data for statistical analysis. How do they treat missing data and bad survey data? And how do they treat nominal, ordinal, interval, and ratio data?
3. Validating Market Segments Using Internal Segmentation Study

If the service provider uses only survey data to develop the market segmentation, make sure they append descriptors such as Prizm segment codes that can be linked back to customers in the database. Otherwise, the segmentation study will become meaningless.

To overcome the limitations of using only primary research data for market segment study, I recommend the analytics team within an organization independently developing market segments using internal data only. This internal market segmentation study involves the following major steps:

• Preparing the data;
• Creating multidimensional segmentation variables;
• Using factor analysis to group like variables if necessary;
• Using cluster analysis to segment customers (i.e., K-Means, Twostep, or Latent Class);
• Developing customer profiles to gain deeper insights into each market segment;
• Using the GE matrix to select the target segment;
• Developing a perceptual map to position a strategy.

After completing the internal segmentation work, compare and cross-check the results with the segments developed by the service provider. These two different segmentation studies will validate the accuracy of each other and shed more light on hidden threats and new growth opportunities, helping the consulting firm to refine their work further and develop a more accurate and comprehensive segmentation scheme.

The beauty of a market segment study is that it will discover risks and opportunities internal data alone cannot otherwise identify. For instance, the vice president of marketing of a diamond fashion jewelry company believes that her primary customers are female self-purchasers. What the research turned out was that male gift-givers account for 40 percent of total jewelry sales. Another example is that you might think skincare and cosmetics products are mainly for female customers, but a segmentation study shows that many male customers also bought skincare products, presenting a promising growth opportunity for the company.

Conclusion

Do not do segmentation just for the sake of segmentation. Segments must be useful and actionable. Smarter marketers make things simple but not simpler. While the one-for-all segmentation strategy seems to be able to simplify things, it will cause more confusion and result in lower ROI of marketing initiatives. Since different departments and different teams will have different needs and are facing various challenges, organizations must build multiple types of segmentation and multiple levels of segmentation to help address their challenges accordingly and effectively.

---