DOES OUTBOUND MARKETING ENHANCE CUSTOMER RETENTION?

(1)

1

DOES OUTBOUND MARKETING ENHANCE

CUSTOMER RETENTION?

What are the moderating effects of relationship length, NPS

promoters, and the number of products on the relationship

between outbound marketing and customer retention?

ROBERT PAUL KORTENOEVER LEPE

University of Groningen

(2)

2

PREFACE

This thesis is the final project of my master in Marketing Intelligence (MI) at the

(6)

6

ABSTRACT

This research aims to better understand how outbound marketing activities can be used to maximize customer retention or, in other words, to minimize customer churn. More specifically, this research seeks to demonstrate whether the frequency of outbound marketing activities, in the form of commercial emails and mails, have an effect on the continuance of relationship between the company and the customer and whether this effect is higher for (1)long term relationship customers, (2) customers who are NPS promoters (i.e. customer who provided a grade of 9-10 in the NPS), (3) heavy usage customers.

This study follows a quantitative approach based on the analysis of a Dutch insurance firm’s customer database. Several logistic regressions are performed to test for the effects of (a) direct mail and email marketing on customer retention as well as for the effect of the following customer loyalty/relationship moderators (b): relationship length, the number of products, and NPS promoters. Overall, our results demonstrate a significant positive relationship between the use of email marketing and the probability of retaining customers. Direct mail, on the other hand, is proven not to have any statistically significant relationship with customer retention. In

(7)

7

(8)

8

CHAPTER 1. INTRODUCTION

In today’s world, there is an upgoing wave of digitalization significantly affecting the marketing strategy of most organizations. There is rising attention towards the generation of online content, the management of social networks, and more specifically, the so-called Inbound marketing. Inbound Marketing can be described as permission-based marketing that aims to earn someone’s attention through social media and engaging content (Rancati et al., 2015). On the other side, we find its counterpart Outbound Marketing, which can be described as the interruption-marketing technique that companies use to attract a broad audience by sending a message publicly through cold calling, direct (e-)mail, paid advertisement trade shows (seminars), etc. ((Dakouan et al., 2019; Hawlk, 2018). In the current age of digitalization, the term “outbound marketing” is considered an old-fashioned approach, and its importance is often ignored. Yet, its potential should not be underestimated. One of its key advantages is that

outbound marketing activities enable reaching certain groups of customers who are not regulars of social networks and online content, as the only way to get your message across these groups is through the use of conventional means such as postal mail or telephone. Outbound marketing activities present additional advantages: (1) is direct - communication is clear on what you have to do: “click here”, “buy”, etc., the marketing message is oriented towards the product or service. (2) consumers are familiar with it, i.e. TV ads, direct mailing; (3) it can deliver results in a short period; (4) it can be effective to increase brand awareness, i.e. TV ads can reach a wider

(9)

9

commercial messages as there are multiple tools than can block the ads. Furthermore, according to Kim et al. (2012) and Wang and Petrison (1993) commercial emails and phone calls can raise the privacy concerns of customers. Moreover, excessive direct mailing can have a negative influence on customers’ attitudes towards direct marketing (Akaah et al., 1995).

Give the pros and cons, we can describe outbound marketing as a “double-edged sword”. On the one hand, it can help increase revenues and sales. On the other hand, customers might feel annoyed by unnecessary and/or unwanted commercial messages. Hence the existing dilemma on whether the use of outbound marketing activities should be seen as a strategic advantage or a source of potential risk.

(10)

10

We have the following research objectives. First, we aim to find to what extent outbound marketing - represented by email marketing and direct mail- influences customer retention. Second, we examine whether the relationship between email marketing/ direct mail is dependent on the relationship length/ NPS promoters/ number of products.

Considering all of these, our main guiding question would be: “Does outbound marketing enhance customer retention?”

This question will be followed by a sub-question: “What are the moderating effects of relationship length, NPS promoters, and the number of products on the relationship between outbound marketing and customer retention?”

To answer both questions a dataset from a Dutch insurance firm’s customer database will be used. Such a dataset includes anonymized contact details, products purchased by customers, socio-demographics, NPS, and whether the customer churned or not.

(11)

11

CHAPTER 2. LITERATURE REVIEW

2.1 Customer retention

Customer retention has been the subject of many studies. The term “Customer Retention” is defined by Gupta and Zeithaml (2006) as “the probability of a customer being alive or repeat buying from a firm”. Another definition provided by Buchanan and Gillies (1990) describes customer retention as “the percentage of customers at the beginning of the year that remains at the end of the year”. The application of this term can differ across industries, e.g. for an Internet services provider customer retention would be defined as the continuity of the relationship with the same service provider; for discount retailers, it would be the continued repeat buying with the retailer (Keiningham, et al. 2007). In our study, given the characteristics of our dataset, retention is not associated with “staying with the company” but “staying with a product acquired” at the insurance service provider. We will explain this more in detail in Chapter 3.

In today’s highly competitive market, retaining customers is of crucial importance for any company’s survival, especially since the loss of a customer has a bigger financial impact than the loss of the next customer acquisition (Alshurideh, 2016; Gupta et al., 2004). As the cost of retaining customers is much lower than the cost of acquiring new customers, customer

(12)

12

2.2 Customer churn

Contrary to customer retention, we define customer churn as “the probability the

customer leaves at a given period” or “as the percentage of the firm’s customer base that leaves in a given period” (Leeflang et al., 2015). Moreover, churn in a contractual setting can be defined as “the termination of the contract between the company and its customer” (Leeflang et al., 2015). In our study, we see customer churn as the process in which customers are terminating the contract with at least one of the products acquired at the insurance service provider.

Managing customer churn plays an important role in Customer Relationship

Management. A high rate of customer churn can derive in severe negative long-term financial consequences (Ascarza et al., 2016). These consequences are translated into an instant loss of sales revenues, an increase in the cost of acquiring new customers, and/or the loss of future cash-flows (Risselada et al., 2010; Ascarza et al., 2016; Gupta et al., 2004).

An important step in preventing customer churn is to offer customers attractive benefits that reduce their willingness to leave or switch to a competitor (e.g. pricing) (Ascarza et al., 2016). These benefits sometimes come in the form of commercial messages that seek to create new sales through up- and cross-selling, i.e. discounts, special offers. Such economic benefits are considered a potential strategy in retaining customers, and the use of different multiple channels to communicate these benefits helps companies to build long-term relationships (Ahmad and Buttle, 2001; Rangaswamy and Van Bruggen, 2005).

One of the most frequent forms of promoting such benefits is through outbound

(13)

13

years. Outbound marketing strategies focus on short-term and delivering high results (Bleoju et al., 2016) and it involves pushing people towards the product or the firm, to obtain an effective response and generate increased sales and revenues (Dakouan et al., 2019).

2.3 Outbound marketing

Outbound marketing is described by Rancati et al. (2015) as the traditional form of advertising, in which people must stop what they are doing to pay attention to the commercial message or deal with it in some other way. The marketing message is one-way from the company to the customer and not vice versa.

Over the years, several models have been developed to maximize customer’s

responsiveness and the company’s expected profit through the execution of outbound marketing activities. One of the most important is the RFM model, based on Recency (i.e. how long it has been since a customer made the last purchase), Frequency (i.e. how often a customer purchased in the past over a certain period) and Monetary value (i.e. how much a customer spent with the company) model- the most important one.

(14)

14

According to Verhoef (2003), outbound marketing techniques such as direct mailing focus mainly on cross-selling, and therefore, it is not expected to have an impact on customer retention. However, recent research has shown that outbound marketing techniques do influence customer retention. Chittenden and Rettie (2003) proved that e-mail marketing is more effective when aimed at customer retention instead of customer acquisition. Furthermore, Kim et al. (2012) suggest that when customers perceive the frequency of e-mails and sales calls exceeds their expectations, they are more likely to end the relationship with the company. Following this line of thought, we believe there is reliable evidence that the frequency of outbound marketing has an impact on customer retention.

Therefore, the starting point of this study considers that outbound marketing activities can affect customer retention in both positive and negative ways.

On the one hand, outbound marketing activities can help customers to acquire knowledge about the company’s attributes, products, discounts, events, and so on. Moreover, effective cross-selling of products and/or services has a positive influence on customer retention since the switching costs are higher when cross-buying increases (Kumar et al., 2008). Furthermore, De Wulf et al. (2001) states that a higher level of direct mail can increase the perceived level of relationship investment. Last but not least, Akaah et al. (1995) show that consumers with a positive past direct marketing experience a tendency to manifest positive direct marketing attitudes.

(15)

15

customer’s expected interval has a negative influence on the customer’s willingness to remain in the relationship with the company. Given these findings, it seems logical to think that customers might feel annoyed by unnecessary and unwanted commercial messages from their insurance service provider, and therefore, we also expect that a high frequency of outbound marketing contact moments can influence negatively the probability of customers staying with the insurance service provider.

The question that remains open is how exactly outbound marketing influences customer retention. Probably the most accurate answer is that outbound marketing can be a “double-edged sword”. Managers should use outbound marketing to generate additional sales, but at the same time, they should be careful with excessive commercial messages because it can irritate customers and increase their willingness to end their relationship with their customers.

An important factor in the relationship between outbound marketing and the customer is the type of activity. The nature of outbound marketing is rather diverse, as it includes a great variety of activities such as direct mail, telemarketing, TV commercials, cold-calling, email marketing, etc. Hence, each activity may impact the customer differently. A relevant aspect to be considered here is the degree of intrusiveness of each activity. Previous studies have

(16)

16

Furthermore, another important factor is how the customer perceives the outbound marketing activity. A defining characteristic of human nature is that the perceptions and

behaviors differ from one individual to another. As a result, not all customers will have a positive towards a certain ad and vice versa. Thus, the relationship between outbound marketing and customer retention will highly depend on the customer itself.

In our study, we focus on two types of outbound marketing activities: direct mail and email marketing. In addition, we also incorporate three variables that accommodates the customer factor previously mentioned. In sections 2.4 and 2.5, we will discuss Direct mail and Email marketing. Moreover, in sections 2.6, 2.7, and 2.8 we will describe three

loyalty/relationship characteristics - relationship length, NPS promoters, and the number of products- which will account for the differences in the customer base.

2.4 Email marketing

Email marketing is one of the most useful and effective tools for promotion today in today’s highly competitive marketplace. It can be described as the act of “directly market a commercial message to people by email” (Wu et al., 2018) or “as the way to promote any business online by sending emails to current or potential customers” (Budac, 2016; Chaffey et al., 2009). In addition, it has been demonstrated that email marketing provides an opportunity to enhance brand loyalty of already loyal customers at a significant low cost (Merisavo and Raulas, 2004) Last, but not least, the financial cost of sending a large number of messages through email is much lower than regular mail (Merisavo and Raulas, 2004).

(17)

17

thought, we formulate our expectation of email marketing affecting customer retention in a positive way as our first hypothesis.

H1: Email marketing has a positive influence on customer retention

2.4 Direct mail

Direct mail in the literature is defined as “an addressed, written, commercial message that is delivered at the addresses by postal service” (Vriens et al., 1998). The goal of this medium is to communicate with potential buyers and increase customer’s purchasing frequency, both up- and cross-sell (Bult and Wansbeek, 1995; Merisavo and Raulas, 2004). Moreover, direct mail provides rewards to customers such as price discounts and helps to raise awareness of products and services (Huang, 2015).

Direct mailing presents several advantages such as precision in targeting (potential) customers, the opportunity to deliver personalized offers, and high flexibility in the format used, timing, and testing (Vriens et al, 1998). Furthermore, it involves no direct competition for the attention of the customer from other advertisements (Verhoef, 2003). On the side, one drawback of direct mail is its considerable high cost per message compared to other tools like e-mailing (Vriens et al., 1998).

As highlighted before, Verhoef (2003) suggests that direct mail does not affect customer retention because it is intended to create additional sales. However, more recent studies

(18)

18

mail on customer retention, led to our first hypothesis on direct mail positively influencing customer retention.

H2: Direct mail has a positive influence on customer retention

2.6 Relationship Length

Relationship length reflects the number of days that an individual has been a customer at the company. Prior studies confirm that relationship length has a negative effect on switching intentions and lapsing rate (Lopez et al., 2006; Bhattacharya, 1998). In addition, according to Balaji (2015), as the relationship length grows customers’ knowledge about the firm increases simultaneously and, therefore, customers are more confident when they evaluate the firm’s offerings and initiatives (Palmatier et al., 2006). Similarly, Kim et al. (2012) state that customers with more experience with a firm have a greater trust (and fewer concerns) in the company that customers with less experience. Based on these pieces of evidence, we expect that the effects of direct mail and email marketing on customer retention will be stronger as relationship length increases.

H3a: The effect of Email marketing on customer retention will be stronger as relationship length increases

(19)

19

2.7 NPS Promoters

The NPS is a customer metric introduced by Reichheld (2003) to measure customer loyalty by asking customers to what extent they would recommend the given company using a Likert scale of 1-10. Depending on the submitted scores, customers can be grouped into

“promoters” (9-10), “passively satisfied” (7-8), and “detractors” (0-6). The NPS is calculated as the ratio of promoters to detractors.

In our investigation, we are only interested in those customers who are “promoters”, as we identify such as a proxy for loyal customers. Given the previously stated benefit of email marketing as a medium to enhance brand loyalty of already loyal customers (Merisavo and Raulas, 2004), we hypothesize that email marketing has a stronger effect on promoters than other types of customers.

H4a: The effect of Email marketing on customer retention will be stronger on promoters

Following this line of thought, we hypothesize that the same effect holds true for direct mail.

H4b: The effect of Direct Mail on customer retention will be stronger on promoters

2.8 Number of products

(20)

20

define the number of products a customer as an indicator of the relationship of the customer with the company as well for customer retention.

Given the strength of this variable as a moderator, we expect that the effect of both direct mail and email marketing on customer retention becomes stronger as the number of products a customer owns increases.

H5a: The effect of Email marketing on customer retention will be stronger as the number of products increases

H5b: The effect of Direct Mail on customer retention will be stronger as the number of products increases

2.9 Conceptual Model

(21)

21 Figure 1. Conceptual Model

The hypotheses H1 and H2 will answer the main guiding question: “Does outbound marketing enhance customer retention?”. Two outbound marketing activities are selected: 1) Email marketing in the form of commercial emails; and (2) Direct mail in the form of

commercial mails. Customer retention will be represented by a binary choice: Stay (1) and Churn (0). A further explanation of the variables Commercial emails, Commercial mails, and Stay-Churn will be provided in Chapter 3.

(22)

22

(23)

23

CHAPTER 3. RESEARCH DESIGN AND METHODOLOGY

3.1 Data overview

The data used for this study was provided by a Dutch insurance company and includes anonymized customer information for the period between June 1st 2015 until 29th April 2019. Such customer information includes anonymized contact details, products purchased by customers, socio-demographics, NPS activity, and whether the customer churned or not.

3.2 Sample selection

To analyze the effects of outbound marketing activities on customer retention we decided to use quarterly data (3 months interval). The selection criterion is that we are interested in the short-term /medium-term effects of direct mail and email marketing rather than the long-term effects (i.e. 1 year, 2 years). Also, we were not interested in analyzing the effects in a more granular way (i.e. 1 month, 1 week) because there is not enough variation. It should be pointed out that this selection responds to our criteria of how to assess direct mail and email marketing campaigns. In practice, most commercial messages have an immediate effect (same week), or a lag effect (2 weeks, 3 weeks, etc.). Therefore, we decided to test for an interval of 3 months, as we can ensure enough variation and a medium/short-term effect.

(24)

24

who do not churn have more emails because they stayed for a longer period, as well as customers who do churn, have fewer emails because they stayed for a shorter period.

For our research, we chose Q3 of 2016 to calculate the frequency of commercial e-(mails) and we validated its effects on Q4 2016. The selection criterion is that Q3 2016 presented a considerable high number of commercial (e)mails compared to other quarters. Hence, it makes it way more interesting to see what are the effects of an increase in the number of commercial messages on customer retention.

(25)

(26)

26

The selection process in Figure 2 can be explained as follows. First, for the "Selection of customers”, only customers who did not churn before the start of the validation period: Q4 2016 were selected. Thus, customers who churned in Q3 (or earlier) are not considered for two main reasons: (1) we will get unreliable results, as the churners in Q3 (or earlier) will have less commercial (e)mails, (2) we want to measure the effects on customer retention on Q4, so it does not make sense to take into account customers who churned before.

Second, we assessed the number of outbound marketing activities that customers received in Q3 2016. Similarly, for the control variables (age and gender) and the moderators (relationship length, NPS promoters, and the number of products) we use the data of Q3 2016. Therefore, all the IVS will have data from Q3 2016.

Finally, we will investigate the effects of the IVS on customer retention in Q4 2016.

3.3 Variables

3.3.1 Dependent variable (DV)

3.3.1.1 Customer Retention - Stay

(27)

27

Moreover, as highlighted in the previous section 3.2 Sample selection, customer retention is evaluated in the 4th Quarter of 2016 (Q+1). Below, in Table 2, a summary of the variable is provided.

3.3.2 Independent variables (IVs)

3.3.2.1 Direct mail – Commercial mails

In our investigation, direct mail is represented by the numeric variable Commercial mails, which indicates the number of commercial postal mails sent to a customer during Q3 2016. Moreover, the details of the content/nature of the commercial mails are not provided due to privacy reasons. Below, in Table 3, the minimum and maximum value, the median, the mean, and the variance of Commercial Mails are presented.

3.3.2.2 Email marketing – Commercial emails

(28)

28

3.3.3 Moderators

3.3.3.1 Relationship length

Relationship length is a numeric variable that reflects the number of years that an individual has remained as a customer with the company before Q3 2016. Below, in Table 5, a summary of Relationship length is provided.

3.3.3.2 NPS Promoters

NPS promoters is a binary variable that indicates whether a customer is an NPS promoter (1) or not (0). NPS promoter (=1) can be described as a customer who filled in an NPS survey within the Q3 2016 and submitted a score of 9-10. Below, in Table 6, a summary of NPS promoters is presented.

3.3.3.3 Number of products

(29)

29

3.3.4 Control Variables

3.3.4.1 Gender

The variable Gender, binary in nature, indicates whether a customer is a male (0) or female (1). A summary of the variable is provided below.

3.3.4.2 Age

The variable Age (in years), discrete in nature, indicates the age of the customer. In Table 9 a summary of the variable Age is provided.

3.3.5 Overview of the variables

Variables Type Description

Stay Binary Stay (1), Churn (0)

(30)

30

Age Numeric Age of the customer in years at the start of Q3

Gender Binary Male (0), Female (1)

Relationship Length Numeric Number of years as a customer at the start of Q3 NPS Promoters Binary Promoter Yes (1), Promoter No (0)

Number of products Numeric Number of products of the customer at the start of Q3 Table 10. Overview of the variables

3.4 Data exploration

3.4.1 Data description. Missing data & outliers

The dataset used for this study includes data from 5 separate datasets: one dataset containing contact details between the customer and the company, one with churn data, one with sociodemographic details, one with product information, and one with NPS activity. The created dataset contains 11339 observations of customers (one for each customer) and 9 columns (1 column that indicates the ID of the customer, 1 column that corresponds to the dependent variable (Stay), and 7 columns that relate to the 7 explanatory variables).

(31)

31

commercial mails in an interval of three months. Furthermore, the variable Number of products presented a few outliers. We investigated each case individually, and we did not find any corrupted records. We concluded that these observations are realistic.

3.4.2 Correlation matrix

(32)

32

3.5 Logit Model

The method of analysis chosen is a binary logistic regression (Logit Model).

A defining characteristic of our DV Stay is that we have two outcomes (1 or 0, Stay or Churn). Therefore, in order to determine the effect of direct mail and email marketing on customer retention, a simple linear regression cannot be applied, and thus, a binary choice model must be selected. In our investigation, a Logit Model will be performed. Our motivation for this model is that the Logit Model is often preferred to Probit Model due to mathematical convenience and easier interpretation of the parameters (Leeflang et al., 2015).

We assume that the error term, i.e. the variance in our dependent variable that is not explained with the explanatory variables in our model, is independently distributed according to the Cumulative Distribution Function (CDF). The logit model is formulated as follows:

𝑃[𝑌 = 1] = Ʌ(𝛽0+ 𝑋′𝑖𝛽)

where the probability of observing Y = 1 is equal to the cumulative distribution function of the error term evaluated at (𝛽₀+ 𝑋′

𝑖𝛽)

Moreover, the cumulative distribution of the error term follows a logistic distribution. We can express our logistic regression as follows:

Ʌ(𝛽₀+ 𝑋′ 𝑖𝛽) =

exp( 𝛽0+ 𝑋′𝑖𝛽)

(33)

33

where 𝛽₀ represents the intercept or constant, 𝑋_𝑖 is the matrix of explanatory variables for consumer i, and 𝛽 is a vector of parameters (Leeflang et al., 2015)

3.6 Moderation analysis

Another important step in our analysis is to investigate the moderating effects of Relationship Length., NPS promoters, and Number of products.

(34)

34

Figure 3. Graphical representation moderation analysis

The moderation analysis in a logit model can be interpreted as follows.

log [ 𝑌 1 − 𝑌] = 𝑎 + 𝑏𝑋 + 𝑐𝑍 + 𝑑(𝑋 ∗ 𝑍) where a is the constant b the effect of X in Y c the effect of Z in Y

(35)

35

3.6.1 Mean centering

To enhance the interpretability of the parameters of the moderation analysis we will use mean-centering. As we are interested in the main effects of X and Z (i.e. the effect b of X on Y when Z=0; the effect c of Z on Y when X=0), we can not interpret i.e. Relationship Lengh=0 and therefore, we need to rescale the variables by subtracting the mean. We are going to adjust our variables un such a way that the transformed variables when they take a value of 0, they will be at the average level.

Interpretations are now relative to the mean. So, if the transformed variable takes a value of 1, it means “one more than the mean” of the old variable

3.7 Plan of analysis

The analysis is structured as follows. First, we will investigate the effects of direct mail and email marketing on customer retention with logistic regression (section 3.7.1). Then, we will test the moderator effects of relationship length, NPS promoters and number of products on the relationship between direct mail and email marketing on customer retention (section 3.7.2)

3.7.1 Do Commercial (e-)mails influence customer retention? (Hypothesis 1 and 2)

In this section, a logit model (Model 1) is performed to investigate the effect of outbound marketing frequency on customer retention.

Model 1 – Main effects Commercial (e)mails

(36)

36 The specification of the model is as follows. 𝑃[𝑌 = 1]

= exp(𝛽0+ 𝛽1𝐶𝑜𝑚𝑚𝑒𝑟𝑐𝑖𝑎𝑙_𝑒𝑚𝑎𝑖𝑙𝑠 + 𝛽2𝐶𝑜𝑚𝑚𝑒𝑟𝑐𝑖𝑎𝑙_𝑚𝑎𝑖𝑙𝑠 + 𝛽3𝐴𝑔𝑒 + 𝛽4𝐺𝑒𝑛𝑑𝑒𝑟) 1 + exp(𝛽0+ 𝛽1𝐶𝑜𝑚𝑚𝑒𝑟𝑐𝑖𝑎𝑙_𝑒𝑚𝑎𝑖𝑙𝑠 + 𝛽2𝐶𝑜𝑚𝑚𝑒𝑟𝑐𝑖𝑎𝑙_𝑚𝑎𝑖𝑙𝑠 + 𝛽3𝐴𝑔𝑒 + 𝛽4𝐺𝑒𝑛𝑑𝑒𝑟)

where

𝛽₀ is the constant

𝛽₁ is the coefficient/ estimate of Commercial emails 𝛽2 is the coefficient/ estimate of Commercial mails

𝛽₃ is the estimate of Age 𝛽₄ is the estimate of Gender

3.7.2 Do Relationship Length/ NPS Promoters/ Number of products strengthen the effect of Commercial (e)mails on customer retention? (Hypothesis 3a, 3b, 4a, 4b, 5a, 5b)

In order to analyze the role of each of the moderators on the effect of direct mail and email marketing on customer retention, a logit model (Model 2) will be performed.

Model 2- Moderation effects

(37)

37

Commercial (e)mails and Relationship Length, Commercial (e)mails and NPS Promoters, Commercial (e)mails and Number of products.

We will use the mean-centered version of the variables Commercial Emails, Commercial Mails, Age, Relationship Length, Number of products. To do so, we will subtract the mean for each of the variables, as described in the previous section.

The specification of the model is as follows. 𝑃[𝑌 = 1] = exp (𝛽0+ 𝛽1𝐸 + 𝛽2𝑀 + 𝛽3𝐴 + 𝛽4𝐺 + 𝛽5𝑅 + 𝛽6𝑁 + 𝛽7𝑃 + 𝛽8 (𝐸 𝑥 𝑅) + 𝛽9 (𝑀 𝑥 𝑅) + 𝛽10 (𝐸 𝑥 𝑁) + +𝛽11(𝑀 𝑥 𝑁) + 𝛽12 (𝐸 𝑥 𝑃) + 𝛽13 (𝑀𝑥 𝑃) ) 1 + exp (𝛽0+ 𝛽1𝐸 + 𝛽2𝑀 + 𝛽3𝐴 + 𝛽4𝐺 + 𝛽5𝑅 + 𝛽6𝑁 + 𝛽7𝑃 + 𝛽8 (𝐸 𝑥 𝑅) + 𝛽9 (𝑀 𝑥 𝑅) + 𝛽10 (𝐸 𝑥 𝑁) + +𝛽11(𝑀 𝑥 𝑁) + 𝛽12 (𝐸 𝑥 𝑃) + 𝛽13 (𝑀𝑥 𝑃) ) where

E is the mean-centered version of Commercial emails

M is the mean-centered version of Commercial mails

A is the mean-centered version of Age

G is Gender

R is the mean-centered version of Relationship Length

N is NPS promoters

(38)

38

𝛽₁ is the coefficient/ estimate of the mean-centered version of Commercial emails 𝛽₂ is the coefficient/ estimate of the mean-centered version of Commercial mails 𝛽₃ is the estimate of the mean-centered version of Age

𝛽4 is the estimate of Gender

𝛽₅ is the estimate of the mean-centered version of Relationship Length 𝛽₆ is the estimate of NPS Promoters

𝛽₇ is the estimate of the mean-centered version of Number of Products

𝛽8 is the estimate of the interaction between the mean-centered version of Commercial

emails and the mean-centered version of Relationship Length

𝛽₉ is the estimate of the interaction between the mean-centered version of Commercial mails and the mean-centered version of Relationship Length

emails and NPS Promoters

𝛽₁₁ is the estimate of the interaction between the mean-centered version of Commercial mails and NPS Promoters

𝛽₁₂ is the estimate of the interaction between the mean-centered version of Commercial emails and the mean-centered version of Number of products

(39)

39

3.8 Multicollinearity

An important step in our analysis is to check whether the models discussed in the previous section presents multicollinearity. Multicollinearity appears when there is a strong resemblance between two independent variables, so we can’t distinguish what causes a change in our dependent variable, customer retention. This makes the parameters estimates unreliable (Leeflang et al., 2015). In order to test multicollinearity, the Variance Inflation Factors (VIF) were calculated, as shown in Tables 12-13-14. Model 1 did not show any symptoms of multicollinearity, as the VIF values (Table 12) didn’t exceed the threshold of 4 (moderate multicollinearity) or 10 (strong collinearity). On the other side, for Model 2 we checked for multicollinearity for the old version of the variables (without mean-centering). In this case, multiple VIF values of Model 2 (Table 13) exceed the threshold of 4 (Commercial emails, Commercial mails, and interaction terms - Commercial mails x Relationship Length and Commercial mails x Number of products) and the threshold of 10 (Relationship Length, NPS Promoters, Number of products and the interaction terms- Commercial emails x Relationship Length, Commercial emails x NPS Promoters and Commercial emails x Number of products), thus showing symptoms of moderate and strong multicollinearity. These high VIF values can be explained by the fact that interaction terms tend to be collinear with the original variables involved. In any case, multicollinearity is considered irrelevant for moderation tests (McClelland et al., 2017), and therefore, we can ignore it.

(40)

40

low (11336 zeros and 3 positive values) (see Table 15). A rule of thumb is that we need at least 5 observations to estimate parameter reliability. Thus, we decided to not include the interaction term Commercial Mails x NPS Promoters in our Model 2 because of the lack of activity (<5 obs).

(41)

41

0 11336 obs.

1 1 obs.

2 2 obs.

Table 15. Activity Interaction term: Commercial Mails x NPS Promoters

CHAPTER 4. RESULTS

4.1 Do Commercial (e-)mails influence customer retention? (H1 and H2)

Variable Estimate Std Error Z-value Pr (>|z|)

(42)

42 a. Interpretation of the significance levels

The output of the variables used in the logistic regression is presented in Table 16. As indicated by the significance levels, the variables Commercial mails and Gender are not significant in determining the probability of a customer “staying” (p>0.05). On the other side, variables representing the age of the customer (Age) and email marketing (Commercial emails) remain significant at a confidence level of 95%.

Given the present results, we can confirm that there is a significant relationship between email marketing and customer retention. On the other hand, Commercial mails seemingly does not influence customer retention, and therefore, we cannot demonstrate a significant relationship between direct mail and customer retention.

b. Interpretation of the effects of the IVS

Coefficient Odds Marginal effects

Commercial emails 0.064 1.066 0.001

Age 0.026 1.026 0.0005

Table 17. Coefficients, odds and marginal effect of Commercial emails and Age

Interpretation of Email marketing effects on Customer Retention

In order to assess the impact of the Commercial emails on customer retention 3 criteria will be used: the coefficients, the odds-ratios, and the marginal effects (see Table 17)

(43)

43

effect of the IV in the probability of observing a customer “staying” (Y=1). Given the results in Table 17, Commercial emails has a positive and incremental effect on the probability of a customer staying (at a 95% confidence level).

The odds ratio is an important criterion when interpreting the parameters of a logit model (Leeflang et al., 2015). The odds can be interpreted as the probability of a customer “staying” (Y=1) relative to the probability of a customer “not staying” (Y=0) (ceteris paribus). The odds ratio is computed with the exponent of the coefficient (beta):

𝑂𝑑𝑑𝑠 𝑟𝑎𝑡𝑖𝑜_𝑘 = exp(𝛽_𝑘)

𝑂𝑑𝑑𝑠 𝑟𝑎𝑡𝑖𝑜_{𝐶𝑜𝑚𝑚𝑒𝑟𝑐𝑖𝑎𝑙 𝐸𝑚𝑎𝑖𝑙𝑠}= exp(𝛽₁) = exp (0.064)= 1.066

Moreover, a change in the odds rate can be interpreted as follows. One unit increase in the explanatory variable changes the odds ratio by exp(𝛽1𝑥 1) = exp(0.064) = 1.066. The odds ratio

Table 17 shows that an increase of one unit in the number of Commercial emails will lead to an increase in the odds of a customer “staying” (Y=1) by a factor of 1.066 (6.6%) (ceteris paribus); two units increase in the number of Commercial emails will lead to an increase the odds of a customer “staying” (Y=1) by a factor of 1.137 (13.7%) (ceteris paribus); etc.

The last criterion, the marginal effects, indicates by how much does the probability of observing a customer “staying” (Y=1) increase, if:

- A binary IV changes from 0 to 1

(44)

44

The marginal effects are calculated using the derivative of the probability with respect to an explanatory variable.

𝜕𝑃 [𝑌𝑖=1]

𝜕𝑋𝑖𝑘 = 𝑃[𝑌𝑖 = 1](1 − 𝑃[𝑌𝑖 = 1])𝛽𝑘

Marginal effects for Commercial emails= 𝑃[𝑌_𝑖 = 1](1 − 𝑃[𝑌_𝑖 = 1])𝛽₁ = 𝑃[𝑌_𝑖 = 1](1 − 𝑃[𝑌_𝑖 = 1])0.064

In our case, an instantaneous change of one unit increase in the number of Commercial emails increases the probability that a customer “stays” (Y= 1) by 0.1% (ceteris paribus); an instantaneous change of two units increase in the number of Commercial emails increases the probability that a customer “stays” (Y= 1) by 0.2% (ceteris paribus); etc. It is worth noting that the marginal effects differ at other points of the data, thus implying that there is no “the marginal effect” but “a marginal effect” at a specific value. In our case, the marginal effects are calculated at the average value/observation.

Similarly, our results can be also interpreted on the probability that a customer churns (Y=0). The number of Commercial emails has a significant negative effect on churn. Furthermore, one unit increase of Commercial emails decreases the odds that a customer churns (Y=0) by a factor of 1.066 (6.6%) (ceteris paribus). Finally, regarding the marginal effects, an instantaneous change of one unit increase in Commercial emails leads to a decrease in the probability of churn (Y=0) by 0.1% (ceteris paribus).

Interpretation of Age (control) effects on Customer Retention

(45)

45 Calculation of the odds and marginal effects: 𝑂𝑑𝑑𝑠 𝑟𝑎𝑡𝑖𝑜𝐴𝑔𝑒 = exp(𝛽3) = exp (0.026)

Marginal effects for Age= 𝑃[𝑌_𝑖 = 1](1 − 𝑃[𝑌_𝑖 = 1])𝛽₃ = 𝑃[𝑌_𝑖 = 1](1 − 𝑃[𝑌_𝑖 = 1])0.026

Based on the results in Table 17, Age has a positive and incremental effect on the probability of a customer staying (at a 95% confidence level). Furthermore, a customer becoming one year older increases the odds that a customer stays (Y=1) by a factor of 1.026 (2.6%) (ceteris paribus). Finally, regarding the marginal effects, an instantaneous change of one year more leads to an increase in the probability of staying (Y=1) by 0.05% (ceteris paribus).

c. Model fit and significance

McFadden Cox & Snell Nagelkerke

0.024 0.005 0.005

Table 18. Pseudo R-squared metrics of Model 1

(46)

46 .

Likelihood Ratio Test Df Log-Likelihood P-value

Null Model 1 -1216.0

Model 1 5 -1187.2 9.48e-12***

Table 19. Likelihood Ratio Test

Furthermore, in order to test if the notation as a whole is significant, we used a Likelihood Ratio Test to compare our model with the null model. The null hypothesis is that “there are no differences between models”. If the p-value is lower than the critical value (0.05) the Null Hypothesis can be rejected. Based on our results (see Table 19), we can conclude that our model is significant (p-value below 0.05), and therefore we can explain significantly more than a model with no variables.

4.2 Do Relationship Length/ NPS Promoters/ Number of products strengthen

the effect of Commercial (e)mails on customer retention? (H3a, H3b. H4a, H4b,

H5a, H5b)

Variable Estimate Std. Error Z-value Pr(>|z|)

Constant 3.921 0.086 45.429 <2e-16***

(47)

(48)

48 a. Interpretation of the significance levels

The resulting outcomes of the moderation analysis are presented in Table 20. As indicated by the significance levels, the moderation effects of Relationship Length, NPS promoters, and Number of products on the relationship between Commercial emails and Commercial mails are not significant (p-value <0.05). On the other side, only the main effects of Commercial emails, Relationship Length and Age are significant in determining the probability of a customer staying (p-value <0.05).

Based on the present results, we are not able to demonstrate any of the moderating effects previously discussed in Chapter 2. The customer loyalty/relationship variables: Relationship Length, NPS Promoters, and Number of products do not strengthen or weaken the effect of Commercial (e)mails on customer retention.

b. Interpretation of the effects of the IVS

Coefficient Odds Marginal effects

Commercial emails 0.052 1.053 0.001

Age 0.017 1.017 0.0003

Relationship Length 0.023 1.024 0.0005

Table 21. Coefficients, odds and marginal effects of Commercial emails, Age and Relationship Length

Interpretation of Email marketing effects on Customer Retention

(49)

49

Commercial emails has a positive and incremental effect on the probability of a customer staying (at a 95% confidence level). Moreover, a change in the odds rate can be interpreted as follows. One unit increase in the explanatory variable changes the odds ratio by exp(𝛽₁𝑥 1) = exp(0.052) = 1.053. The odds ratio Table 21 shows that an increase of one unit in the number of Commercial emails (one more than the average level) will lead to an increase the odds of a customer “staying” (Y=1) by a factor of 1.053 (5.3%) (ceteris paribus); two units increase in the number of Commercial emails (two more than the average level) will lead to an increase the odds of a customer “staying” (Y=1) by a factor of 1.110 (11.1%) (ceteris paribus); etc. The last criterion, the marginal effects, indicates that an instantaneous change of one unit increase in the number of Commercial emails (one more than the average level) increases the probability that a customer “stays” (Y= 1) by 0.1% (ceteris paribus); an instantaneous change of two units increase in the number of Commercial emails (two more than the average level) increases the probability that a customer “stays” (Y= 1) by 0.2% (ceteris paribus); etc.

Similarly, our results are also applicable to the probability that a customer churns (Y=0). The number of Commercial emails has a significant negative effect on churn. Furthermore, one unit increase of Commercial emails (one more than the average level) decreases the odds that a customer churns (Y=0) by a factor of 1.053 (5.3%) (ceteris paribus). Finally, regarding the marginal effects, an instantaneous change of one unit increase in Commercial emails (one more than the average level) leads to a decrease in the probability of churn (Y=0) by 0.1% (ceteris paribus).

(50)

50

The interpretation of both Age and Relationship Length will be relative to the mean because they are mean-centered.

The output in Table 21, shows that Age and Relationship Length have a positive and incremental effect on the probability of a customer staying (at a 95% confidence level). Furthermore, a customer becoming one year older (compared to the average level) increases the odds that a customer stays (Y=1) by a factor of 1.017 (1.7%) (ceteris paribus) whereas a customer being one more year at the company (compared to the average level) increases the odds that a customer stays (Y=1) by a factor of 1.024 (2.4%) (ceteris paribus). Finally, regarding the marginal effects, an instantaneous change of one year older (than the average level) leads to an increase in the probability of staying (Y=1) by 0.03% (ceteris paribus), whereas an instantaneous change of one year more at the company (than the average level) leads to an increase in the probability of staying (Y=1) by 0.05% (ceteris paribus).

c. Model fit and significance

To assess the fit of the logit model we use again the so-called pseudo R-squared metrics: Mcfadden, Cox & Snell, and Nagelkerke. The values are shown in Table 22.

McFadden Cox & Snell Nagelkerke

0.03 0.006 0.006

(51)

51

The Mcfadden 𝑅2_{of our model has a value of 0.03 while Cox & Snell and Nagelkerke}

have a value of 0.006. The values slightly improved compared to Model 1. Likelihood Ratio Test Df Log-Likelihood P-value

Null Model 1 -1216.0

Model 1 5 -1187.2 1.323e-10***

Table 23. Likelihood Ratio Test

Furthermore, in order to test if the notation as a whole is significant, we used a Likelihood Ratio Test to compare our model with the null model. The null hypothesis is that “there are no differences between models”. If the p-value is lower than the critical value (0.05) the Null Hypothesis can be rejected. Based on our results (see Table 23), we can conclude that our model is significant (p-value below 0.05), and therefore we can explain significantly more than a model with no variables.

4.3 Overview of the hypotheses

In this section, the hypotheses stated in Chapter 2 will be either accepted or rejected given the previous statistical analysis.

Hypothesis Accepted / Rejected

H1: Email marketing has a positive influence on customer retention Accepted

(52)

52

H3a: The effect of Email marketing on customer retention will be stronger as relationship length increases

Rejected

H3b: The effect of Direct mail on customer retention will be stronger as relationship length increases

Rejected

H4a: The effect of Email marketing on customer retention will be stronger on promoters

Rejected

H4b: The effect of Direct mail on customer retention will be stronger on promoters

Rejected

H5a: The effect of Email marketing on customer retention will be stronger as the number of products increases

Rejected

H5b: The effect of Direct mail on customer retention will be stronger as the number of products increases

(53)

53

CHAPTER 5. CONCLUSION

5.1 General Discussion

The main goal of this study is to provide evidence that outbound marketing activities provide positive effects on customer retention. To do so, we measured the impact of direct mail and email marketing on the probability of a customer staying (Y=1), as described in Chapters 4. Subsequently, our second goal is to demonstrate whether the relationship between email

marketing/ direct mail is dependent on customer loyalty/relationship indicators. Consequently, we analyzed the moderating effects of relationship length, NPS promoters, and the number of products, as described in Chapter 4. Building on the analyses conducted, we can proceed to answer the main research question (i.e. “Does outbound marketing enhance customer retention?) and the sub-question (i.e. “What are the moderating effects of relationship length, NPS

promoters, and the number of products on the relationship between outbound marketing and customer retention?”)

Our first hypothesis (H1: “Email marketing has a positive influence on customer

retention”) was accepted. The significant impact of email marketing on customer retention is in line with Chittenden and Rettie (2003), who consider email marketing a major tool for customer retention. This study suggests that the use of email marketing is more effective in retaining customers than in acquiring customers. Our results show that email marketing in the form of commercial emails enhances customer retention and does not entail a risk source for the company.

(54)

54

studies of Doaei et al. (2011) and Malliari and Sirkeci (2017), whose findings demonstrated a positive effect of direct mail on customer loyalty and on attitudinal loyalty, as well as in trust and relational commitment, our statistical analysis shows that direct mail seemingly does not influence customer retention. This supports the assumption hold by Verhoef (2003), which states that direct mail’s only purpose is to create additional sales and therefore, it does not influence customer retention.

Based on these two hypotheses, we can conclude that outbound marketing does enhance customer retention, yet the magnitude of such influence greatly depends on the type of activity. More specifically, we observe that direct mail does not influence the decision to stay/continue with products of the company whereas email marketing does influence the decision to stay/continue with the products of the company. Hence, only one outbound marketing activity has been proven to affect customer retention.

Our third hypotheses (H3a/b: The effect of Email marketing/Direct mail on customer retention will be stronger as relationship length increases) were rejected. The relationship between email marketing/direct mail is not dependent on the number of days a customer has been with the company. There is no difference between the old customers/ new customers. However, Relationship Length does have a positive effect on customer retention, which is line with Lopez et al. (2006) and Bhattacharya (1998), as they showed that relationship length had a negative effect switching intentions and lapsing rate.

(55)

55

brand loyalty of already loyal customers, we found no distinction between loyal/nonloyal customers. On the other hand, we were not able to analyze the effect of promoters on the relationship between direct mail and customer retention due to a lack of activity (only 3 customers -out of 11339- were promoters and received at least one commercial mail). Moreover, the effect of NPS promoters on customer retention was not significant, thus implying that whether a customer is a promoter or not seemingly does not matter for retaining customers.

Our fifth hypotheses (H5a/b: The effect of Email marketing/Direct mail on customer retention will be stronger as the number of products increases) were not accepted. The number of products seemingly does not influence the relationship between email marketing/ direct mail and customer retention. In addition, the number of products does not have any significant effect on the probability of retaining customers contradicts the findings of Donkers et al. (2003), which suggests that having more products will reduce the probability that customers churn.

Based on the outcomes of our hypotheses, we can affirm that the moderating effects of relationship length, NPS promoters, and the number of products on the relationship between outbound marketing activities are not significant. Hence, there is no difference between customers with better/higher customer characteristics and customers with worse/lower customer characteristics.

5.2 Theoretical implications

(56)

56

in order to maximize future purchases. Through this study, we demonstrate a positive effect of one activity of outbound marketing- email marketing- on customer retention. Even though the main purpose of email marketing, and more specifically, commercial emails, remains to be the generation of more sales, we have shown that they can help retain customers. On the other side, we have shown that direct mail is not relevant for customer retention purposes, following the line of thought of Verhoef (2003).

Furthermore, as a result of our statistical analyses, we have shown that the age of the customer and the number of days a customer has been with the company have a significant influence on customer retention. Both variables are proven to be important in predicting customer retention/churn.

Moreover, our moderation analyses suggest that the customer loyalty/relationship indicators: relationship length, NPS promoters, and the number of products are not relevant for retention purposes on email campaigns/ direct mail campaigns.

To conclude, we can affirm that both direct mail and email marketing are not risk factors for the company (from retention’s perspective). As we already discussed before, outbound marketing can negatively affect the company’s reputation (Rancati et al., 2015) as well as excessive outbound marketing activity can increase churn intention (Kim et al., 2012). Our results demonstrate that direct mail and email marketing do not increase customer churn.

5.3 Managerial implications

(57)

57

considered more than just an activity to increase sales. As highlighted in previous chapters, commercial emails help to retain customers. Besides, email marketing is a better medium than other traditional channels. The cost of sending emails is much lower than regular posts (Merisavo and Raulas. 2004; Chittenden and Rettie, 2003) and is less intrusive than other outbound marketing activities such as telephone calls. Apart from that, it is of crucial importance to not send excessive commercial activities, as suggested by Kim et al. (2012). Companies must ensure appropriate implementation of such commercial messages to minimize negative attitudes towards such emails and the company. Moreover, we encourage managers to monitor the retention rate and the response rate of the emails regularly. Keeping track of the performance of emails over time can be very effective in the long run. Furthermore, we suggest putting effort into the designing of the commercial email Even though we did not incorporate the content of the email on our research, we expect that it would have an impact on the effectiveness of the email. Finally, we recommend using A/B testing to test the different designs of the commercial emails, and therefore, selecting the best choice.

5.4 Limitations and future research

(58)

58

(59)

59

REFERENCES

Ahmad, R., & Buttle, F. (2001). Customer retention: a potentially potent marketing management strategy. Journal of strategic marketing, 9(1), 29-45.

Akaah, I. P., Korgaonkar, P. K., & Lund, D. (1995). Direct marketing attitudes. Journal of Business Research, 34(3), 211-219

Alshurideh, M. T. (2016). Is customer retention beneficial for customers: A conceptual background. Journal of Research in Marketing, 5(3), 382-389.

Ascarza, E., Iyengar, R., & Schleicher, M. (2016). The perils of proactive churn prevention using plan recommendations: Evidence from a field experiment. Journal of Marketing Research, 53(1), 46-60.

Balaji, M. S. (2015). Investing in customer loyalty: the moderating role of relational characteristics. Service Business, 9(1), 17-40.

Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of personality and social psychology, 51(6), 1173.

Bhattacharya, C. B. (1998). When customers are members: Customer retention in paid membership contexts. Journal of the academy of marketing science, 26(1), 31-44.

(60)

60

Buchanan, R. W., & Gillies, C. S. (1990). Value managed relationships: the key to customer retention and profitability. European Management Journal, 8(4), 523-526.

Bult, J. R., & Wansbeek, T. (1995). Optimal selection for direct mail. Marketing Science, 14(4), 378-394.

Budac, C. (2016). Theoretical approaches on successful email marketing campaigns. Ovidius University Annals, Economic Sciences Series, 16(2), 306-311.

Chaffey, D., Ellis-Chadwick, F., Mayer, R., & Johnston, K. (2009). Internet marketing: strategy, implementation and practice. Pearson Education.

Chittenden, L., & Rettie, R. (2003). An evaluation of e-mail marketing and factors affecting response. Journal of Targeting, Measurement and Analysis for Marketing, 11(3), 203-217.

Dakouan, C., Benabdelouahed, R., & Anabir, H. (2019). Inbound marketing vs. outbound marketing: independent or complementary strategies. Expert Journal of Marketing, 7(1).

De Wulf, K., Odekerken-Schröder, G., & Iacobucci, D. (2001). Investments in consumer relationships: A cross-country and cross-industry exploration. Journal of marketing, 65(4), 33-50. Doaei, H., Rezaei, A., & Khajei, R. (2011). The impact of relationship marketing tactics on customer loyalty: the mediation role of relationship quality. International Journal of Business Administration, 2(3), 83.

Donkers, B., Franses, P. H., & Verhoef, P. C. (2003). Selective sampling for binary choice models. Journal of Marketing Research, 40(4), 492-497.

(61)

61

Gupta, S., & Zeithaml, V. (2006). Customer metrics and their impact on financial performance. Marketing science, 25(6), 718-739.

Hawlk, K. (2018). Outbound Versus Inbound Marketing: Which Strategy Is Best?. Journal of Financial Planning, 31(6), 30-31.

Huang, M. H. (2015). The influence of relationship marketing investments on customer gratitude in retailing. Journal of Business Research, 68(6), 1318-1323.

Keiningham, T. L., Cooil, B., Aksoy, L., Andreassen, T. W., & Weiner, J. (2007). The value of different customer satisfaction and loyalty metrics in predicting customer retention, recommendation, and share‐of‐wallet. Managing service quality: An international Journal.

Kim, M., Park, J. E., Dubinsky, A. J., & Chaiy, S. (2012). Frequency of CRM implementation activities: a customer‐centric view. Journal of Services Marketing.

Kumar, V., George, M., & Pancras, J. (2008). Cross-buying in retailing: Drivers and consequences. Journal of Retailing, 84(1), 15-27.

Leeflang, P., Wieringa, J. E., Bijmolt, T. H., & Pauwels, K. H. (2015). Modeling markets. Springer-Verlag New York.

Li, H., Edwards, S. M., & Lee, J. H. (2002). Measuring the intrusiveness of advertisements: Scale development and validation. Journal of advertising, 31(2), 37-47.

(62)

62

Malliari, L., & Sirkeci, I. (2017). Performance of direct mail in building customer loyalty in Greek automotive sector. Int. J. of Business Performance Management, 18(1), 1-24.

McClelland, G. H., Irwin, J. R., Disatnik, D., & Sivan, L. (2017). Multicollinearity is a red herring in the search for moderator variables: A guide to interpreting moderated multiple regression models and a critique of Iacobucci, Schneider, Popovich, and Bakamitsos (2016). Behavior research methods, 49(1), 394-402.

Narayandas, D. (2005). Building loyalty in business markets. harvard business review, 83(9), 131-139.

Merisavo, M., & Raulas, M. (2004). The impact of e‐mail marketing on brand loyalty. Journal of Product & Brand Management.

Palmatier, R. W., Dant, R. P., Grewal, D., & Evans, K. R. (2006). Factors influencing the effectiveness of relationship marketing: A meta-analysis. Journal of marketing, 70(4), 136-153.

Piersma, N., & Jonker, J. J. (2004). Determining the optimal direct mailing frequency. European Journal of Operational Research, 158(1), 173-182.

Rancati, E., Codignola, F., & Capatina, A. (2015). Inbound and Outbound marketing techniques: A comparison between italian and Romanian pure players and click and mortar companies. Risk in Contemporary Economy, 2(1), 232-238.

(63)

63

Reichheld, F. F. (2003). The one number you need to grow. Harvard business review, 81(12), 46-55.

Rejón-Guardia, F., & Martínez-López, F. J. (2014). Online advertising intrusiveness and consumers’ avoidance behaviors. In Handbook of strategic e-business management (pp. 565-586). Springer, Berlin, Heidelberg.

Risselada, H., Verhoef, P. C., & Bijmolt, T. H. (2010). Staying power of churn prediction models. Journal of Interactive Marketing, 24(3), 198-208.

Van Doorn, J., & Hoekstra, J. C. (2013). Customization of online advertising: The role of intrusiveness. Marketing Letters, 24(4), 339-351.

Verhoef, P. C. (2003). Understanding the effect of customer relationship management efforts on customer retention and customer share development. Journal of marketing, 67(4), 30-45.

Vriens, M., Van der Scheer, H. R., Hoekstra, J. C., & Bult, J. R. (1998). Conjoint experiments for direct mail response optimization. European journal of Marketing.

(64)

64

Appendix A- Descriptive Statistics

A.1 Boxplots of explanatory variables

(65)

65

(66)

66 98% 2% 0% 20% 40% 60% 80% 100% Stay Churn

Female population

98% 2% 0% 20% 40% 60% 80% 100% Stay Churn

Male population

A3. Distribution of Stay/Churn on Gender

(67)

67

APPENDIX B. R-code

rm(list=ls()) #set wd setwd("Y:/staff/bew/M&M/Other/masterthesis_data") #read files contact_details=read.csv("contact_details.csv") demo=read.csv("demo.csv") klantbatch_id=read.csv("klantbatch_id.csv") products=read.csv("productbezit.csv") nps=read.csv("nps.csv") indep=read.csv("zoekdata.csv") ########## DATA PREPARATION ########################

#1# Change dagen to date

contact_details$dagen = as.Date(contact_details$dagen, origin = "2015-06-01") contact_details$year = format(as.Date(contact_details$dagen,

format="%d/%m/%Y"),"%Y")

contact_details$q <- quarters(as.Date(contact_details$dagen))

klantbatch_id$dagen_regend[klantbatch_id$dagen_regend == 2916307] <- 1428 klantbatch_id$dagen_regend = as.Date(klantbatch_id$dagen_regend, origin = "2015-06-01")

(68)

68

klantbatch_id$year<- format(as.Date(klantbatch_id$dagen_regend, format="%d/%m/%Y"),"%Y")

#2# Selection of customers

#interested in customers from Q4 2016 (validation period) klantbatch_id$dummy <- rep(0,nrow(klantbatch_id)) klantbatch_id$dummy[klantbatch_id$dagen_regend >= "2016-10-01"] <- 1 klantbatch_id <- klantbatch_id[klantbatch_id$dummy == 1,] klantbatch_id$stay <- rep(0,nrow(klantbatch_id)) klantbatch_id$stay[klantbatch_id$dagen_regend >= "2017-1-01"] <- 1 klantbatch_id$stay <- as.factor(klantbatch_id$stay) #3# OUTBOUND MARKETING library(dplyr) #year 2016 df2016 <- contact_details[contact_details$year == "2016",] #q1# #eMail outbound df2016q1 <- df2016[df2016$q == "Q1",] #email subset

email2016 <- df2016q1[df2016q1$bron == "EM",] #Email outbound

email2016q1 <- email2016[email2016$inuit == "O",]

(69)

69

freq16q1 <- commercialemail2016q1 %>% group_by(klant_id,omschrijving_1) %>% summarise(Freq=n()) mean(freq16q1$Freq) #q2# #eMail outbound df2016q2 <- df2016[df2016$q == "Q2",] #email subset

email2016b <- df2016q2[df2016q2$bron == "EM",] #Email outbound

email2016q2 <- email2016b[email2016b$inuit == "O",]

commercialemail2016q2 <- email2016q2[email2016q2$omschrijving_1 == "Commerciële mailing verstuurd",]

freq16q2 <- commercialemail2016q2 %>% group_by(klant_id,omschrijving_1) %>% summarise(Freq=n()) mean(freq16q2$Freq) #q3# #eMail outbound df2016q3 <- df2016[df2016$q == "Q3",] #email subset

email2016c <- df2016q3[df2016q3$bron == "EM",] #Email outbound

email2016q3 <- email2016c[email2016c$inuit == "O",]

commercialemail2016q3 <- email2016q3[email2016q3$omschrijving_1 == "Commerciële mailing verstuurd",]

DOES OUTBOUND MARKETING ENHANCE CUSTOMER RETENTION?