Innovation Diffusion & Marketing:
What is the influence of different marketing mechanisms on the diffusion
process of an innovation?
By
Merel Husken
Faculty of Economics and Business MSc Marketing
Index
Abstract p. 3
Chapter 1: Introduction p. 4
Chapter 2: Literature Review p. 7
2.1 Diffusion of innovations p. 7
2.2 Factors influencing the diffusion curve p. 10
2.2.2 Product p. 10 2.2.3 Promotion p. 11 2.2.4 Word-of-Mouth p. 15 2.3 Conceptual model p. 16 Chapter 3: Methods p. 17 3.1 Data Description p. 17 3.2 Model Development p. 20 3.3 Analysis Procedure p. 22 3.4 Assumptions p. 23 Chapter 4: Results p. 25
4.1 Outcomes from the analysis Petcare p. 25
4.1.1 focusing on the hypothesis p.26
4.2 Outcomes from the analysis Confectionary p. 27 4.2.1 focusing on the hypothesis p. 28
4.3 Outcomes of further analyses for best model fit p. 29
Chapter 5: Discussion p. 33
Chapter 6: Conclusion p. 35
References p. 37
Abstract
Data is increasingly taking a central role in organizations. The amount of data and methods to analyze it increases every day. While the richness and size of modern customer databases offer opportunities for companies, this development also comes with challenges. There is a need for more powerful metrics and analytics that will cause companies to work more effectively and efficiently. Because, companies are continuously making improvements and new products, managers are interested in understanding the sales growth of their innovations, as well as the aspects that affect it. One of the most highly cited papers in the marketing literature is the 1969 Bass Model paper, which focuses on the diffusion of innovations. Within this research four contributions have been made to the original Bass diffusion model. First, a difference will be made between the untapped market, the effective potential market and the current market. Second, a difference is made between non-triers, triers, non-repeat and repeaters. Third, this thesis adds to the literature by including the effect of marketing and price promotions on the diffusion of the innovation. Finally, the analysis is done by splitting the data according to the degree of innovation.
Chapter 1: Introduction
The last several years, a challenge has occurred for many companies. They find challenges in getting useful insights out of big data. It is not clear to managers within organizations which combination of price and marketing mechanisms have an effect on a successful introduction of an innovation. Specifically, how do these mechanisms influence the penetration of the product and each other, and which is most influential in its contribution to a successful introduction. This is partly because they do not have enough time to do big time-consuming analysis, but also because they just do not always know all the opportunities which are available while analyzing the big amount of data flowing in. Whether companies lack the right knowledge, expertise or capabilities, they might be missing out on valuable insights, and maybe even revenue growth, meeting customer needs or targeting the wrong people. A lot of companies and even countries struggle with ways how to use data (Chandy et al., 2017). Not very surprising, because it has been estimated that there will be 20 billion devices connected to the internet of things. This results in big amounts of data (Gartner, 2015). Walmart for example collects 2,5 petabytes (2.500.000 GB) of information every hour (McAfee et al., 2012), and this is just the beginning. Companies are now working to connect offline and online data seamlessly (Hui, Fader and Bradlow, 2009a), run experiments that can combine information from different data sources in real-time (Anderson & Simester, 2003) and use eye tracking data to maintain and build customer relationships (Chandon et al., 2008; Van der Lans, Pieters, and Wedel, 2008). Combining these new methods to the always increasing technical improvements (like IP address tracking, cookies, loyalty cards etc.), will enable companies to collect every moment, every transaction and even every touchpoint of their consumers. Therefore, data is increasingly taking a central role in organizations (Wedel & Kannan, 2016).
While the richness and size of modern customer databases offer opportunities for companies, this development in increasing amounts of customer data also comes with challenges. Some articles state that it is not about getting more data, but about better data (Bradlow et al., 2017). But, according to Bradlow et al. (2017), only acquiring better data is not enough, companies should also have the knowledge and capabilities to be able to get actionable insights out of this data. A lot of companies have focused primarily on investing in data capture and storage, and not enough in analytics. Marketing analytics has become more important than ever. There is a need for more powerful metrics and analytics that will cause companies to work more effectively and efficiently (Wedel & Kannan, 2016).
interested in understanding the sales growth of their innovations, as well as the aspects that affect it (Ruiz-Conde, 2005). According to Ruiz-Conde (2005), this implies understanding the diffusion process of innovations. The diffusion of an innovation is defined as the process through which the innovation “is communicated through certain channels over time among the members of a social system” (Rogers, 1983, p.5). The “Bass Diffusion Model” is used as the basis of this research paper. This model is inspired by the diffusion of innovations of Everett Roger (1962), which classifies adopters of an innovation according to the timing of their adoption. He specifies five different classes: Innovators, Early Adopters, Early Majority, Late majority and Laggards. According to this theory, apart from Innovators, adopters are influenced by imitation in varying degrees: the pressure of the social system and the pressure of increasing later adopters (Bass, 2004). Bass (1969) introduced diffusion models into marketing literature. He states that the adoption process of a product is similar to an epidemic spread. “People who have not adopted the innovation are “infected” by those who have and are influenced by external sources like advertising” (Bass, 1969). Understanding the diffusion process of new products is key in strategic planning of marketing (Ruiz-Conde, 2005).
The 1969 Bass Model paper is one of the most highly cited papers in the marketing literature (Parker, 1994; Bass, 2004), that is why this research uses this model. Furthermore, it has a big acceptance in the field of innovation diffusion (Mahajan, Muller and Wind, 2000). Even though, the Bass Model has been accepted broadly, the model is based on assumptions which limits its applicability. That is why, researchers have made several modifications to the model. An example of this, is that the model initially was intended for consumer durable innovations, while nowadays, it has been proven that its applicable to a lot of other innovations (Ruiz-Conde, 2005).
Ruiz-Conde (2005) states that more research is needed to contribute to the Bass model literature, in order to develop useful tools to understand the diffusion processes of different kinds of innovations in different settings. Jorgensen (1983, p. 269) says: “In a marketing context, the purpose of the model is to describe and predict the increase over time in the number of adopters of a new product. However, from a product manager’s point of view, many diffusion models may not be very useful since they fail to incorporate the firm’s marketing decision variables such as price and advertising”. To contribute to the literature, this research paper builds further on the extension of the Bass model presented in the article of Ruiz-Conde (2005), focusing on price and advertising.
The research question therefore is:
What is the influence of different marketing and price mechanisms on a successful introduction of an innovation?
To answer this question, weekly level sell-out data is used per brand, as well as internal data on the different mechanisms and promotions during the introductions. The program R is used to do the analysis.
In this research paper, price and marketing mechanisms are included in the diffusion model. This research contributes to the gab which exist in the literature, by looking at brand level (Krishnan, Bass & Kumar, 2000) and including advertising (Ruiz-Conde, 2005).
Chapter 2: Literature Review
2.1 Diffusion of innovations
Every year many companies introduce new products to stay competitive in an ever changing market and to increase their long-term performance (Srinivasan et al., 2009). Especially FMCG (fast-moving consumer goods) companies are continuously improving products and bringing innovations to the market. They invest a great amount of recourses to be able to meet customers’ needs (Sinapuelas, Clark & Bohlmann, 2015). But not every innovation is a success. More than 85% of Fast Moving Consumer Goods fail every year (Wengel, 2014). For innovations to be successful, customers first need to adopt the product. According to Roger (1962) diffusion of innovations helps to understand how, why and how fast new products and ideas spread. Rogers classifies 5 groups of consumers based on their adoption speed. The 5 classes he specifies are: Innovators, Early Adopters, Early Majority, Late majority and Laggards.
The innovators are the first 2,5% of people which will adopt the innovation. These are people who like to explore new ideas and products, and are willing to take risks. They follow trends and are very interested in innovations. The early adopters are the following 13,5% of the people who adopt the innovation. These people are the “opinion leaders”. They have a high social status and are able to influence and inform people about the product. These consumers are more discreet about their choices than the innovators. The early majority are the next 34% and are called the “followers”. These consumers will be inspired by the opinion leaders and will get informed about the products through others (e.g. reviews). The late majority is the 34% of the consumers which are sceptical at first but still want to fit in. They often have the fear of missing out. They do not easily switch to the innovation and are elaborate in their research. Lastly, the Laggards are the last 16% of the consumers to adopt. These people prefer the traditional products and will only adopt the innovation if there is no alternative available or if they feel pressure from others (Rogers, 1962; Rogers, 1976; Rogers, 1983).
So, looking at customers during an innovation, overall we find 2 types of customers: innovators and imitators. Innovators are the consumers who adopt the innovations before the rest of the market. Innovators are more externally influenced. This externally influenced effect is also called the advertising effect (Bass, 1969). The other group is called imitators. Imitators are the followers. These consumers will adopt the innovation as a result of others adopting it. They are “infected” by other adopters. Imitators are more internally influenced, also called the word-of-mouth effect (Bass, 1969).
According to the literature, one of the main challenges for managers responsible for the introduction of new products, has been to get a better understanding of the mechanisms influencing the adoption of innovations (Guo, 2008; Rogers, 2003). Currently, there is a growing body of research in marketing and innovation, showing that social contagion plays a crucial role in the diffusion of innovations (Bell & Song, 2007; Manchanda et al. 2008). Social contagion is “the process by which consumers influence each other to adopt and use a product in a specific way” (Langley et al., 2012).
Social contagion has been proven to play a big role in the diffusion of adoption of new products. In the literature the assumption is made that social contagion is caused by consumer characteristics and by the interactions with others around them (Langley et al., 2012). This phenomenon can work through explicit recommendations (e.g. WOM), implicit social norms (e.g. feeling about expectations) or by visible behavior (e.g. using the product or seeing others purchasing the product)(Burt, 1987; Manchanda et al., 2008; Langley et al., 2012; Van den Bulte & Stremersch, 2004). Next to these, several papers have provided different theoretical accounts of social contagion, which include social learning, social-normative pressures, competitive concerns and network effects (Van den Bulte & Lilien, 2001).
There are several different theories explaining social contagion or the so called infection that is passed on from one consumer to the next. Below, we discuss three of them. The first theory is called the “fear of missing out” (FOMO). This theory suggests people tend to have a negative feeling when others are doing of having something that person is not part of. Przybylski define FOMO as “a pervasive apprehension that others might be having rewarding experiences from which one is absent” (Przybylski et al., 2013). This phenomenon may cause customers to buy a product they see used by others (Iyengar & Lepper, 2000).
The second theory that might explain social contagion is normative conformity. Normative conformity is defined as “matching one’s behaviour to the responses of others with the goal of obtaining social approval” (Cialdini & Goldstein, 2004). This could lead to consumers choosing the same product as one’s friends in order to fit in the group. People change their behavior, because they want to match to the responses of others. Cialdini & Goldstein (2004) state that people want to have meaningful relationship with each other, and tend to adapt their choices to improve their relationships. They also found that people look at social norms to effectively respond to social situations.
is for people to comply with request of others (Cialdini &Trost, 1998). So, close friends and family have an even stronger influence on the behavior of people.
Overall, the explanation for the S-shaped innovations diffusion curve has long been dominated by the social contagion theory (Van den Bulte & Stremersch, 2004). This S-shaped innovation diffusion curve is the cumulative of the diffusion of innovation bell Rogers (1962) introduced. The first part of the graph is the start of the diffusion. This is the first period after the introduction. This shows low increase in adopters, because only the innovators will try the product. The second part of the S-curve is the part where the “imitation” begins, here you see a high increase in adopters, followed by a stagnating line which means the potential market has been reached.
The Bass diffusion model belongs to the category of contagion models. As previously touched upon, the Bass model is an effective forecasting tool for the diffusion of innovations. The diffusion of an innovation is defined as “the process through which the innovation is communicated through certain channels over time among the members of a social system” (Rogers, 1983, p.5). Earlier, diffusion was compared to the spread of an epidemic. This means that the success of a diffusion is closely linked with the way it spreads through a social system. A failed diffusion does not necessarily imply that the innovation was adopted by no one. A failed diffusion implies that a diffusion does not get the adoption of the potential market that was predicted. The potential market is “the upper limit of the total number of adoptions that are possible” (Ruiz-Conde, 2005). This segment contains the total number of potential customers of the innovations in the social system. If an adoption is not diffused properly, this can be caused by several reasons such as, no complete product, high competition, or not enough awareness (Moore, 1991).
In a seminal paper, Bass (1969) introduced the diffusion models into marketing literature. According to the model, people who did not adopt the new product are “infected” by the other consumers who have, and are influenced by external sources like advertising (Bass, 1969). Frank Bass stated it like this: “The probability that an initial purchase will be made at T given that no purchase has yet been made is a linear function of the number of previous adopters” (Frank M. Bass)
This model has the form:
› n(t): number of adopters at time t › N(t): total number of adopters until t › M: potential market size
› p: innovation parameter › q: imitation parameter
Bass makes a distinction between innovators and imitators. The effect of innovators are displayed within the model as innovation parameter p, which gives the probability of adoptions due to the effect of advertisements (Bertotti et al., 2016). Next, the effect of imitators is displayed within the model as imitation parameter q. The imitation parameter q times the number of adopters before t divided by the potential market size, gives the probability of adoptions due to the effect of word-of-mouth (Bertotti et al., 2016).
2.2 Factors influencing the diffusion curve
Within this research paper, several contributions are made to the Bass diffusion model. The factors influencing the diffusion curve are Product, Promotion and Word-Of-Mouth (WOM). In the next paragraphs these factors contributing to the change in the model are explained in further detail.
2.2.1 Product
Next to different customer groups, it is important to look at the degree of innovations as well. Studies show that there are differences between incremental innovations and radical innovations (Dewar & Dutton, 1986). A study from Osinga, Leeflang and Wieringa (2010) even suggests there should be more research done on the effect of innovativeness on why one brand exhibits more persistent effects than others. That is why the degree of innovativeness of an innovation is also considered in this thesis. The more innovative a new product is, the more risk and uncertainty is associated with the purchase of that product (Emami & Dimov, 2017; Wärneryd 1988; Wolff 2007). The degree of innovativeness also decreases the rate of adoption (Rogers, 2003). There are several studies which highlight the importance of the degree of innovations in adoption processes (Kahn, 2006; Henderson & Clark, 1990; Abernathy & Clark; 1988). Kahn (2006) even identifies seven different types of new products according to the level of newness: Zero newness, compatibility, cost reductions, product improvements, line extensions, new generation products, and new-to-the-world.
product such as “better flavor” or “new and improved”. These are changes that enhance the current product to better fit the customer needs. The fifth type of new products, line extensions, has a higher level of newness that the previous groups. These products are substantially different from extension products, but stays within the line or brand guidelines of the company. The sixth type of new products are called new generations products. These are products that are different in every aspect of the product. This group is a radical innovation. Lastly, the seventh type of new product are the new-to-the-world products. This group scores the highest in newness. These products have never been seen before and are, as the name says, new to the world. These products come with a lot of uncertainty and high levels of risk (Kahn, 2006; Emami & Dimov, 2007; Carayannopoulos 2009; Henderson and Clark 1990).
In this thesis the innovations are classified in the basis of the scale Kahn introduced ranging from incremental innovations, such as product improvements and line extensions, until radical innovations. In the next paragraph the second contribution to the Bass diffusion model is explained in further detail: Promotion.
2.2.2 Promotion
Before adopting a new product, customers go through an adoption process. The customer adoption process is “the mental process a person goes through from first learning about an innovation to the actual adoption of the product”(Kotler & Armstrong, 2016). A consumer goes through 5 different stages when adopting a new product: Awareness, interest, evaluation, trial and finally adoption. These stages are crucial for a person to go through, before they decide to repeatedly buy the innovation. The first stage “awareness” is the process though which a person is hearing about the new product, learning about its existence and becoming aware. Within this stage the customer knows very little about the product and is not really giving the product much attention. This stage leads to the second stage “interest”. After a customer is aware of the new product, they begin to actively search for information. This information can come from advertising for example, but also from current users. After this stage, the third stage “evaluation” starts. With all the information the customer has collected, he is now forming his opinion. He is evaluating and comparing information about the quality, performance and even features of the new product. This will lead to the decision to either try the product or not. If a customer has decided to try the product, he enters the “trial” stage. This is the stage where the person will be able to use, touch and smell the innovation himself. And finally, the fifth stage “adoption” is the stage where a customer has decided to buy the product and adopt the innovation (Kotler & Armstrong, 2016).
Fast Moving Consumer Goods (FMCG) and durable goods. FMCG are low involvement goods with low switching costs, while durable goods are high involvement goods with high switching costs. A lot of customers of FMCGs use their first purchase as a trial. While for durable goods, it is not possible to do so. If you buy a car for the first time, you do not just try the car by buying it. So, there is a huge difference between these types of goods. A lot of research has focused on the trial phase, while for FMCGs it can easily be a one-time thing. A Nielsen study found that brands lose around half their customers between the trial purchase and repeat purchase (Malek, 2018; Kim et al., 2020). That is why it is important to not just look at the first purchase, but also take into repeat purchases into consideration, while examining the success of the adoption of the innovation.
Even when a customer enters the evaluation stage, it does not automatically mean they will try the product as well. That is why it is important for companies to know how they can influence the customer at different stages of the adoption process. According to Hahn et al. (1994) sources of indirect product experience can stimulate customers to enter the trial stage. This can be advertising or other marketing efforts. While on the other hand, to stimulate customers to do repeat purchases and to really adopt the innovation, word-of-mouth by customers who already bought the new product is more effective (Hahn et al., 1994).
Using marketing efforts to promote a new product or to increase sales has been proven to be effective. But there are a lot of different mechanisms to use, which are all affecting different parts of purchasing process of a customer. In the following section a distinction is made between different marketing mechanisms.
It is essential for companies to know when different marketing mechanisms are most effective during the diffusion of an innovation as well. Consumers are exposed to hundreds of marketing messages a day, from Billboards to online advertising, and from TV ads to displays, every brand has its own agenda (American Marketing Association, 2018). This means that not every well-thought-out brand message will get inside the mind of your customer. A famous quote pins the point: “Half the money I spend on marketing is wasted; the trouble is I don't know which half”. This begs the question: What to do?
“One of the most important marketing activities to accompany the product’s introduction is advertising” (Horsky & Simon, 1983). This is because advertising can be used to inform innovators. Moreover, advertising will let the innovators know the new product exists, while at the same time a company can make claims about its quality. Innovators are willing to adopt an innovation in favor of a delay to find out more “experience-based” information about its quality e.g. through friends. Innovators will accept the information the company gives, such as advertising and in-store displays.
Therefore, companies invest a lot of resources in long-term marketing efforts, like brand advertising or out-of-store advertising focused on brand improvement and awareness. This is important because studies show that out-of-store advertising has primarily an effect on the attitude towards a brand, while brand advertising has found to increase awareness and knowledge of the product (Steenkamp & Gielens, 2003; Kim et al., 2020). Moreover, brand advertising increases recall and has a positive effect on purchase intention (Till & Baack, 2005). These marketing efforts are not focused on direct results such as a peak in sales, but are used to make a connection with customers and enhance their attitudes about the brand or product (Smith & Yang, 2004). Using these different mechanisms is key to the long-term value of a brand. It creates loyalty, attachment and even increases the perceived quality of a brand or product (Buil et al., 2013). Furthermore, it can reduce the perceived risk associated with a new product (Baack et al., 2016). Overall, brand and out-of-store advertising can increase sale in the long-run (Jurca & Madlberger, 2015).
Furthermore, FMCG companies often use in-store marketing efforts such as feature advertising and in-store displays. The positive effects of these marketing efforts have been found in several different articles (e.g. Ballings, McCullough, & Bharadwaj, 2018; Steenkamp & Gielens, 2003; Sinapuelas et al., 2015). “Feature advertising refers to advertising of products in newspaper inserts and store flyers. In-store display refers to setting up items in special stacked configurations of at the end of an aisle” (Kim et al., 2020). These in-store marketing efforts promote the new product at the place where customers make their purchases. This way they will be able to influence their behavior ((Berk, Mela, & Van Heerde, 2008; Steenkamp & Gielens, 2003). Studies show that the hedonic value created by the in-store environment has a high influence on the time spend, as well as the amount spend in store (Fam, Merrilees, Richard, Jozsa, Li & Krisjanous, 2011; Mohan, Sivakumaran & Sharma, 2013). It is valuable to create a suitable environment and to be present where customers make their purchases, because studies show that two thirds of the product & brand decisions are made in the store (Fam et al., 2011; Jantarat & Shannon, 2016; Sigurdsson et al., 2010; Mohan et al., 2013), and are likely to be influenced by factors within the store environment like displays or feature advertising (Sigurdsson et al., 2010).
innovators as possible about the existence of the new product. This will result in innovators adopting the product, which in turn will provide information to the rest of the market: the imitators. The company can reduce the advertising when the majority of the innovators has been reached, because imitators are not as much influenced by these advertisements as the innovators are (Horsky & Simon, 1983). Leading to the first and second hypothesis:
H1: Out-of-store promotion and brand advertising are more effective for innovators than imitators in making trail purchases.
H2: In-store marketing efforts are more effective for innovators than imitators in making trail purchases.
Next to instore promotion and advertising, FMCG companies invest heavily in price promotions. Price promotions are proven to increase the sales, because these efforts improve the utility of a consumer by reducing the sacrifice associated with the purchase (Guha et al., 2018). Thus, the greater the customer perceives the discount the more sales it generates (Urbany et al., 1988). Companies have been creative utilizing this technique, from inflating reference prices to stacking discounts (Tuttle, 2016; Chen & Rao, 2007). However, this may not be the best strategy at hand. Price promotions are often used as a means to attract customers to a store and generate more traffic, but cannot attract customers to a store on regular basis (Grewal, Monroe, & Krisbnan, 1998; Lichtenstein & Bearden. 1989). Furthermore, while it might attract customers for a short peak in sales, it can have a negative effect on the perceived quality of the brand, as well as customers internal reference price for that product (Grewal et al., 1998).
Price promotion might not cause sustainable growth, and cannot be used to increase sales on a regular basis, but it can change customer purchase decisions (Innman & McAlister, 1993) and drive customers to try out a new product (Rizwan et al., 2013). So, when an imitator is hesitant about whether or not to buy the innovation for the first time, it is more likely for him to make the purchase when there is a price promotion (Kim et al., 2020). More precisely, studies found that a moderate discount (5-35%) has a positive effect on customer retention, while a low (<5%) and high (>35%) discount has a negative effect on customer retention (Del Rio Olivares et al., 2018). A recent study adds that high promotion has a negative effect on the long run (Kim, 2019). We therefore hypothesize:
H3: Price promotions are more effective for imitators than innovators in making trail purchases.
informal channels can be used, such as participation in workshops, or companies can choose to make use of sampling (Frattini et al.,2014). This way imitators are more informed and certain about the product and its quality, and will be more likely to buy the innovation. This leads to the fourth hypothesis:
H4: Direct product experiences are more effective for imitators than innovators in making trail purchases.
Next to Product and Promotion, there is a third factor influencing the diffusion curve at a later stage in de adoption process: word-of-mouth. In the following paragraph the effect of this factor is explained further.
2.2.3 Word-of-Mouth
If a company has reached the innovators and early adopters, they are up for the next challenge. Because it is very hard to make the step from the innovators to the mass market. This step is defined as “the chasm” by Moore (1991), which has to be crossed in order to make the innovation successful. According to Moore (1991) there are different tactics for crossing the chasm. As a first tactic companies need to create a whole product: a product that is complete and without bugs. Furthermore, the product needs to be positioned appropriately, the price has to be set relative to the competition and the product needs to be distributed through the right channels (Moore, 1991).
But in contrast to the innovators, the imitators will wait for feedback and experiences of the people around them. The likelihood of them to adopt and repeatedly buy the innovation increases when the number of previous adopters increases. This is caused by several reasons: when the number of adopters is growing the minority position of the nonadopters gets more crucial, this will create social pressure. Furthermore, the risks of buying the innovation reduces when more and more people adopt this new product. Imitators have more information about the product and are more certain about its quality (Horsky & Simon, 1983).
Word-of-mouth (WOM) has proven to be an effective method to increase sales. Dost et al. (2019) found that using WOM in marketing campaigns the total sales increase by approximately 3%-18%. Trusov et al. (2009) even concludes that WOM referrals have a substantially longer carry over effect and a higher response elasticity than traditional marketing activities. There is even evidence that WOM has a significant positive effect on brand loyalty (Ngoma, Ntale & Wright, 2019). Thus, contagion can arise because of a combination of indirect and direct influences from prior adopters (Du & Kamakura, 2011). So, to be able to let the innovation diffuse further and generate repeat purchases, the biggest influence is that of word-of-mouth by customers who already bought the new product (Frattini et al.,2014).
H5: Word-of-mouth is more effective during later stages (repeat purchases) of the diffusion process than the initial stages.
2.3 Conceptual model
Figure 2 shows the conceptual model of this research paper. This model gives a clear picture of the relationships between de different variables which are being tested.
Figure 2: Conceptual model
Trail Promotion
Out-of-store & brand advertising In-store displays
Price promotion Direct product experience
Chapter 3: Research design
In this chapter the research design will be discussed. Starting with the description of the data, followed by the model development and lastly the analysis procedure and the assumptions will be discussed.
3.1 Data Description
To research which hypothesis can be confirmed and which cannot, data will be used to do the analysis. This data is weekly sell-out data on brand level within two different categories. The analysis is done in two categories. Confectionary entails chocolate products, while Petcare entails dog and cat food as well as snacks. The comparison is made between an obvious impulse category and a category which 80% of the purchases are included in the shopping list respectively. The data points are reaching from the beginning of the first introduction (April 2018) until the end of 2019.
Within the data there are several introductions which will be the focus of this thesis (table 1 & 2). In the table below, the introductions are shown together with the period they are introduced to the market, their potential market volume and the degree of innovativeness of the introduction.
Table 1: Introductions Confectionary
Product name Month of introduction Potential market volume1
Degree of innovativeness
Introduction 1 April 2018 6000 Line extension
Introduction 2 April 2018 5000 Product improvement
Introduction 3 April 2018 5000 Line extension
Introduction 4 April 2018 8000 New generation
Introduction 5 July 2018 6000 Line extension
Introduction 6 October 2018 4000 New generation
Introduction 7 February 2019 5000 Line extension
Introduction 8 March 2019 6000 New generation
Introduction 9 August 2019 6000 Line extension
Table 2: Introductions Petcare
Product name Month of introduction Potential market volume2
Degree of innovativeness
Introduction 10 April 2018 3000 Line extension
Introduction 11 March 2019 2000 New generation
Introduction 12 April 2018 1000 Line extension
Introduction 13 August 2019 1000 Line extension
Introduction 14 April 2018 4000 Line extension
Introduction 15 June 2019 5000 New generation
Introduction 16 February 2019 2000 Line extension
1 Looking at the base volume of similar items.
The dataset contains information for each product introduction, namely: sales and marketing mechanisms used during and after the introduction. These marketing mechanisms are divided in different categories: price promotions, in-store marketing efforts, out-of-store marketing efforts and advertising, and direct product experiences.
First, price promotions entail discounts like: 1 for …, 2 for …, 20% discount, 25% discount or second for half of the price. Within this research paper, all these price promotions are combined in 1 variable indicating if there was a price promotion (1) or not (0). Second, in-store marketing efforts entails activations like: Display, Extra Goedkoop Droog- en Kruidenierswaarden (EGDKW) and Extra Goedkoop Kassarek (EGKR). Displays are second citing’s in the shop with extra products. EGDKW are stackings of products at the end of an aisle, and EGKR are small baskets near or just before the cashier desk. Third, out-of store marketing efforts and advertising include above-the-line (ATL) advertising like TV campaigns as well as below-the-line (BTL) advertising like online advertising. Finally, direct product experiences are in-store sampling and online sampling.
Table 3 shows an overview of the overall brands descriptive for confectionary, such as how many weeks different marketing mechanisms are used in the time period of the data.
Table 3: Confectionary Overall Promotions Descriptives
N Range Min Max Sum Mean
YearWeek 277 139 201813 201952 Units Current 277 980 13 994 19.163 69 Display 25 0 1 1 EGKR 35 0 1 1 EGDKW 8 0 1 1 PricePromotion 59 0 1 1 ATL 118 0 1 1 BTL 93 0 1 1
go hand in hand with a price promotion. the combination of these two are used in order to get a higher uplift on their promotion.
Additionally, in the graph below (graph 1) the unit sales of the different innovations is shown over time.
Graph 1: Innovations Confectionary over time
Looking at the graph, striking is that some innovations only last a couple of months, while others stay on the market for a longer time. Some innovations are limited editions (only in the market for a limited time frame), while other innovations are intended to stay on the market for a long time. The 2 highest peaking innovations are the limited editions, introduction 5 and 9 Caramel. While they had the same expectations. Furthermore, introduction 4 and 8 is only sold in the summer, so these innovations will drop when the temperature is going down again. This is obviously visible in the graph above. Looking at the other innovations, some just did not reach their potential and were put off the market, while other innovations did well and were officially added to the assortment. Overall, the repeat sales of most innovations do not appear to be the case.
The opposite happens to a lot of the Petcare innovations. The Petcare innovations often show a stable sell-out pattern over time (see graph 2).
Graph 2: Innovations Petcare over time
For cat there are no short term introductions visible in the data range of this paper, but looking at the data for dog, you see two innovation with a high table like peak, but which are delisted in the beginning of 2020.
To be able to analyze this data on the basis of diffusion, a model needs to be developed. In the next paragraph, the model development will be further explained and discussed in detail. 3.2 Model Development
As mentioned before, the Bass diffusion model is used as the basis of the model created in this thesis. There is a wide acceptance of the Bass diffusion model in the literature (Mahajan et al., 2000), but simultaneously researchers are invited to extend the model. The Bass Diffusion model assumes that an adopter will only purchase the innovation once (Ruiz-conde, 2005). While this is applicable to durable goods, it is less suitable for FMCG goods. With these goods people need to make repeat purchases in order to really adopt the product. Additionally, the Bass model does not include the effect of marketing mechanisms on the diffusion of products, while a lot of companies invest a huge amount of money in marketing. Several other researchers already extended the Bass model by including the possibility to analyze repeat purchases as well as the effect of marketing mechanisms (Mahajan et al. 1983; Hahn et al, 1994, Ruiz-Conde, 2005). So, to build further on their efforts the following variables are included in this model.
First, a difference will be made between the untapped market, the effective potential market and the current market. The potential market is “the upper limit of the total number of
0 1.000 2.000 3.000 4.000 5.000 6.000 201814 201817 201820 201823 201826 201829 201832 201835 201838 201841 201844 201847 201850 201901 201904 201907 201910 201913 201916 201919 201922 201925 201928 201931 201934 201937 201940 201943 201946 201949 201952 202003 202006 202009 202012 202015
Introduction 10 introduction 12 Introduction 12 introduction 13
adoptions that are possible” (Ruiz-conde, 2005). This potential market is calculated looking at similar products of the same brand in the current market.
Second, a difference will be made between non-triers, triers, non-repeat and repeaters. Non-triers are the people who are within the potential market but have never tried the innovation. Triers are the consumers who have tried the innovation for the first time. After the consumers try the product, they have the choice to buy it again or not. Consumers who buy the innovation several times after their initial purchase are repeaters, but consumers who decide to never purchase the innovation again are called non-repeaters. Furthermore, within the model the distinction will be made between Innovation and imitation. Thus, the model will divide the customers into 3 groups of adopters, namely: the innovative trail, the imitative trail and the repeat buyers. The first two groups are modelled just like the original Bass diffusion model, with innovators and imitators. Innovators are the first small group of adopters of the product, they are the first ones to discover the innovation, and influence the second group of customers: the imitators (Bass, 1969; Ruiz-Conde, 2005). So, while the innovators are influenced by external communication, imitators are more influenced by internal communication (Rogers, 1961). Following these two groups are the repeat buyers. These adopters are influenced by the interaction between the previous adopters and potential adopters, which is called word-of-mouth (Ruiz-Conde, 2005).
Third, this thesis adds to the literature by including the effect of marketing and price promotions on the diffusion of the innovation. Within the model the effect of marketing mechanisms is measured on 𝛽". 𝛽" is the innovation parameter, which gives the probability of adoptions due to the effect of advertisements. This means the effect of marketing on the innovators is measured. Furthermore, the word-of-mouth effect is measured as the imitation parameter (𝛽#). 𝛽# times the number of adopters before t divided by the potential market size, gives the probability of adoptions due to the effect of word-of-mouth. When we look at the effect of marketing mechanisms on this 𝛽$, we will measure the effect of marketing has on WOM. Moreover, 𝛽$ is the parameter which indicated the repeat sales. In the literature there is found that repeat sales are mostly affected by WOM, that is why this 𝛽$ will not be influenced by marketing variables, but will only focus on the potential sales in post-trial segments.
𝑠&,( = +𝛽",& + . 𝛽"/&
0
/1"
ln4𝑥&/,(6 + 𝛽#&7𝑠&,(8"
𝑚( :; <𝑚(− 𝑞&,(8"? + 𝛽$&𝑞&,(8"
With 𝑞&,(− 𝑞&,(8" = 𝑠&,( − 𝛽$𝑞&,(8"
Where, for product i (i = 1, … , N) in week t (t = 1, … , T):
𝑠&,( = sales of product i in time t (sales from trail and from repeat purchases);
𝑥&/,( = own marketing expenditures on instrument j ( j = 1: Display; j = 2: EGKR; j = 3: EGDWK; j = 4: Price promotion; j = 5: ATL; j = 6: BTL);
𝑚( = total market sales in time period t
𝑞&,(8" = potential sales of product i to customers in post-trial segments (triers, repeaters and buyers of competing brands that have tried product i before) at time t-1;
𝛽",& , 𝛽"/& , 𝛽#& , 𝛽$& = parameters to be estimated.
The level of aggregation is weekly. Using weekly data, the prediction of sales will be accurate and relatively precise, while using monthly data the sales prediction is less precise and to broad. Horvath and Wieringa (2008) found that weekly aggregation level data can be based on the household interpurchase cycle. They also stated that the lower the aggregation level is the better, because this leads to more degrees of freedom. Daily level data is not available, so this is why weekly aggregation level will be used.
3.3 Analysis Procedure
In the first two paragraph’s, the data were described and the model developed. In this paragraph the analysis procedure will be described. The analysis procedure is important to know in order to be able to test your hypothesis and to eventually get the answer to the research question.
To test hypothesis 1 and 2, the effect of out-of-store promotions, brand advertising and in-store marketing efforts are measured on 𝛽"as well as 𝛽#. By comparing these models, the hypothesis can either be confirmed or denied. To see which model is most effective, meaning the marketing efforts have a higher effect on either one of the beta’s, information criteria (AIC & BIC) are used to judge the model fit. The innovation parameter is mostly externally influenced, and these variables will therefore expect to have an higher effect on the innovation parameter than the imitation parameter.
influenced, and therefore these variables will expect to have an higher effect on the imitation parameter than the innovation parameter.
To test hypothesis 5, the effect of WOM has to be measured on repeat sales. There is no direct measure indicating WOM, but the measure for WOM in this model will be indirect in nature. The word-of-mouth effect on sales will result from the 𝛽$ in combination with the potential sales in the post-trail segments. In this thesis there will not be a focus on the fifth hypothesis because of the lack of direct data. Therefore, in the next section we will only focus on the first four hypothesis.
After testing for the hypothesis by comparing the effect of different marketing efforts on the beta’s in the model, further analysis is done. During these analysis different models are compared in order to get the best model possible. A difference is made between the effect of marketing mechanisms on different Beta’s. Additionally, there is a comparison between the two different categories: Petcare and confectionary.
Within this thesis the program R is used to analyze the secondary data. Before the data can be analyzed, the data needs to be prepared. First of all, the descriptives are done in order to get a feeling of the data. The outliers or missings are dealt with. The data contains a lot of NA’s in the marketing variables, indicating there is no mechanism that week for that innovation. To do the analysis, all the NA’s will be replaced with zeros. These zeros have to be numeric in order for it to work, so this is checked and corrected when needed.
The model chosen to do the analysis is the diffusion model indicated in paragraph 3.2. The test which is used to get the results is a complex regression. As mentioned before, there are several parameters in the model which still need to be estimated, namely the Beta’s. But to calculate the Beta’s you need to have q. So, to resolve this issue the first Beta value will be extracted from the literature (Ruiz-Conde, 2005). With this Beta, q can be estimated and with this q the new value of Beta can be calculated. This loop will continue to run until the last value of Beta matches the new value of Beta. This value will be close to the most relevant Beta, and will be used in the final model. This will be done for every Beta in the model and this way the analysis will be made possible.
3.4 Assumptions
There are several assumptions that lie at the basis of diffusion models in general, namely: - the population of potential adopters is limited and remains constant throughout time; - all members of the population eventually adopt the innovation;
- the population mixes homogenously;
- all adopters are imitators and only adopt after having seen others use the innovation; - the rate of adoption does not only depend on the number of previous adopters but
- the probability of two individuals making contact is equal for any two individuals. (Dodds, 1973; Sharif and Ramanathan, 1981)
The assumptions stated above will be used in this thesis as well. To add, there are 9 assumptions Ruiz-Conde (2004) used as well, of which a couple will be relaxed in this paper.
1. The diffusion process is a binary process and population is homogeneous. 2. The size of the adopter population does not change.
3. The parameters of external and internal influence remain constant. 4. Only one adoption per adopter is allowed.
5. Geographical frontiers do not vary. 6. The innovation is diffused in isolation:
7. The characteristics of an innovation and its perception do not alter. 8. There are no supply restrictions.
9. The impact of the marketing variables used to diffuse an innovation is implicitly captured by the model parameters.
Assumption 3 will be relaxed: We assume that the parameters of external and internal influence vary over the diffusion process of the innovation. More explicitly: We assume that marketing variables affect the external and/or internal influence over the diffusion process Assumption 4 will be relaxed: We explicitly incorporate repeat purchases into the diffusion model.
Assumption 9 will be relaxed: We incorporate marketing variables, such as in-store marketing efforts, out-of-store marketing efforts and advertising and direct product experiences (Ruiz-Conde, 2005).
Chapter 4: Results
In chapter 4 the results of the analysis will be discussed. First the outcomes of the analysis and the hypothesis will be explained in further detail and after that the outcome of further analysis for the best model fit will be discussed.
As mentioned earlier, whether or not a hypothesis is confirmed or denied is based on the best model fit, using information criteria (BIC & AIC) to judge whether the variables are more effective on external influence (𝛽") or internal influence (𝛽#). In table 4 (Petcare) and 5 (Confectionary) you see the overview of which hypothesis is confirmed (x), denied (-) or inconclusive (?) according to the information criteria. Moreover, the positive and negative significant variables are displayed in the last 2 columns.
4.1 Outcomes of the analyses Petcare
To start, the first thing that is obvious from table 4, is that 𝛽", (baseline of external influence) and 𝛽$ (repeat purchases) are almost always positively significant in every model. However, the significance of marketing related variables differs across the innovations. While PricePromotion and sampling tend to have a significant positive effect on 𝛽# (imitators), ATL has a negative effect on imitators as well as innovators. Striking is, that Display has a positive effect on imitators but a negative effect on innovators.
Table 4: Overview Petcare hypothesis confirmation
Hypothesis Significant variables Degree of
innovation Name 1 2 3 4 Positive Negative
3. 𝛽",, 𝛽$ 4. 𝛽",, 𝛽#(samplings), 𝛽$ High Introduction 15
-
1. 𝛽",, 𝛽$ 1. 𝛽2. 𝛽# #(ATL) Med Introduction 16- -
1. 𝛽2. 𝛽",, 𝛽$ ",, 𝛽$4.1.1 Focusing on the hypothesis
In this paragraph, the findings will be discussed for every hypothesis.
H1: Out-of-store promotion and brand advertising are more effective for innovators than imitators in making trail purchases.
When comparing the effect of marketing mechanisms on external influence (𝛽",) versus internal influence (𝛽#) according to different information criteria, there can be concluded that the first hypothesis is denied for Petcare cat. As Shown in table 4 neither of the cat innovations can confirm the first hypothesis (Appendix 1). On the other hand, this hypothesis can be confirmed for Petcare dog. Every dog innovation is able to confirm the first hypothesis (Appendix 1). This means that the use of out-of-store marketing efforts for the cat innovations, is more effective on imitators than innovators, while for dog innovations the out-of-store marketing efforts are more effective on innovators than imitators.
H2: In-store marketing efforts are more effective for innovators than imitators in making trail purchases.
Furthermore, when comparing the effect of marketing mechanisms on external influence (𝛽",) versus internal influence (𝛽#) according to different information criteria, there can be concluded that the second hypothesis can be denied for the Petcare cat innovations, but confirmed for the Petcare dog innovation (Appendix 1). Again, the effect for in-store marketing efforts for cat innovations is more effective on imitators than innovators, while for dog innovations in-store marketing efforts are more effective for innovators than imitators. H3: Price promotions are more effective for imitators than innovators in making trail purchases.
H4: Direct product experiences are more effective for imitators than innovators in making trail purchases.
Moreover, when comparing the effect of marketing mechanisms on external influence (𝛽",) versus internal influence (𝛽#) according to different information criteria, there can be concluded that the fourth hypothesis can be confirmed. Almost all Petcare innovations can confirm this hypothesis (Appendix 1), meaning samplings are more effective on imitators than innovators in the diffusion of Petcare innovations.
Overall, what is striking about these results is first, that there are great differences between dog and cat innovations. This results in hypothesis 1 and 2 being confirmed for dog innovations while they are not confirmed for cat innovations. Second, in-store marketing efforts have a greater effect on imitators than innovators, while this is not what was expected. Third, Price Promotions have a bigger effect on innovators but this effect is not consistent over the different products. This might be explained by the fact that there is often only one promotion in a year. Fourth, samplings consistently have a bigger effect on imitators, which is confirmed by the positive significant effect of external influence (𝛽#samplings) on sales. Looking at the differences between products with a high versus low degree of innovativeness there is no distinct pattern found. This emphasizes the heterogeneity of the models.
4.2 Outcomes of the analyses Confectionary
Moving on to table 5, the first thing that is clear, is that 𝛽", (baseline of external influence) and 𝛽$ (repeat purchases) are almost always positively significant in every model. However, the significance of marketing related variables differs across the innovations. Sometimes PricePromotion, BTL and EGDKW are positively significant, while sometimes these turn out to be negatively significant. What is striking is that, every time ATL and Display are significant they have a negative effect on innovators and thus on sales.
Table 5: Overview Confectionary hypothesis confirmation Hypothesis Significant variables Degree of
innovation Name 1 2 3 Positive Negative
High Introduction 4
- - x
1. 𝛽$ 2. 𝛽$, 𝛽#.#(EGDKW) 3. 𝛽$, 𝛽#(PricePromotion) Med Introduction 5- x ?
1. 𝛽", , 𝛽$ 2. 𝛽",, 𝛽$ 3. 𝛽", , 𝛽$ 1. 𝛽#.#(BTL) 2. 𝛽# High Introduction 6x x x
1. 𝛽", , 𝛽$ 2. 𝛽",, 𝛽$ 3. 𝛽$ 1. 𝛽# 2. 𝛽# Med Introduction 7x x x
1. 𝛽",, 𝛽#, 𝛽$ 2. 𝛽", , 𝛽$ 3. 𝛽$ 2. 𝛽"."(Display), 𝛽".#(EGKR) High Introduction 8- - x
1. 𝛽$ 2. 𝛽$, 𝛽#(EGDKW) 3. 𝛽$, 𝛽#(PricePromotion) 1. 𝛽#."(ATL) Med Introduction 9- x x
1. 𝛽",, 𝛽$ 2. 𝛽",, 𝛽$ 3. 𝛽",, 𝛽$ 2. 𝛽".#(EGKR), 𝛽#4.2.1 Focusing on the hypothesis
In this paragraph, the findings will be discussed for every hypothesis.
H1: Out-of-store promotion and brand advertising are more effective for innovators than imitators in making trail purchases.
When comparing the effect of marketing mechanisms on external influence (𝛽",) versus internal influence (𝛽#) according to different information criteria, there can be concluded that the first hypothesis can be denied for confectionary innovations. While four of the innovations confirm the first hypothesis, the other five prove the opposite (Appendix 2). This is why it cannot be confirmed that out-of-store marketing efforts are more effective for innovators than for imitators, or even the other way around.
H2: In-store marketing efforts are more effective for innovators than imitators in making trail purchases.
H3: Price promotions are more effective for imitators than innovators in making trail purchases.
Additionally, when comparing the effect of marketing mechanisms on external influence (𝛽",) versus internal influence (𝛽",) according to different information criteria, there can be concluded that the third hypothesis can be confirmed for the confectionary innovations (Appendix 2). Almost every innovation confirms this hypothesis, mean price promotions are more effective for imitators than for innovators.
Overall, what is striking about these results is first, for the ‘ice’ innovations all marketing efforts are more effective on imitators rather than innovators. This is supported by the positive significant effect of PricePromotion, EGDKW and BTL. Second, overall the out-of-store marketing efforts seem to be more effective on imitators than innovators, while in-store marketing efforts are more effective on innovators. Looking at the differences between products with a high vs low degree of innovativeness there is no distinct pattern found. This emphasizes the heterogeneity of the models.
4.3 Outcomes of further analyses for best model fit
After being able to confirm or deny the proposed hypothesis, the search for the most optimal model proceeds. To be able to predict the sales most accurately, multiple different models have been compared for every innovation. Table 6 shows the models used in the comparison for Petcare. The last column shows the color of the line used in the models comparison figure (figure 2)
Table 6: Petcare models
Model Variables Color
Model 1 ATL and BTL on 𝛽" Red
Model 2 Display, EGDKW on 𝛽", Samplings on 𝛽# Grey
Model 3 PricePromotion on 𝛽" Green
Model 4 PricePromotion on 𝛽", Samplings on 𝛽# Yellow Model 5 Samplings on 𝛽", PricePromotion on 𝛽# Orange Model 6 All variables op 𝛽" except Pricepromotion and sampling Purple Model 7 ATL, BTL and PricePromotion on 𝛽" Black
The results of the analysis are shown in table 7. Additionally figure 2 shows an example for the visualization of the different models of introduction 11. Overall, model 1 has the highest model fit (Appendix 3). This model includes ATL and BTL on 𝛽" (innovators). After the analysis, it has become clear that the models where the effect of the marketing variables are measured on 𝛽" (innovators) are better than on 𝛽# (imitators). Which beta’s are significant is different for every innovation, but 𝛽", (baseline of external influence) and 𝛽$ (repeat) are almost always significantly positive for every model.
Model Significant
Name 1 2 3 4 5 6 7 Positive Negative
Introduction 10
x
𝛽",, 𝛽$ 𝛽# Introduction 11x
𝛽",, 𝛽$ 𝛽""(PricePromotion), 𝛽#(Samplings) Introduction 12x
𝛽$ Introduction 13x x
𝛽",, 𝛽$ 𝛽# Introduction 14x
𝛽",, 𝛽$ Introduction 15x
𝛽",, 𝛽"." (Display), 𝛽#,𝛽$ Introduction 16x
𝛽",, 𝛽$ Figure 2: Model fit introduction 11To conclude, for Petcare innovations the first model is the most effective. This model includes ATL and BTL promotions measured on 𝛽" (innovators) (Appendix 3). Thus, the original model which results from the literature (ATL, BTL, Display, EGKR, EGDKW on 𝛽"and PricePromotions and samplings on 𝛽#), is not the most effective one for Petcare innovations.
Table 8 shows the models used in the comparison for confectionary. The last column shows the color of the line used in the models comparison figure (figure 3)
Model Variables Color
Model 1 ATL and BTL on 𝛽" Red
Model 2 Display, EGKR, EGDKW on 𝛽" Green
Model 4 PricePromotion on 𝛽" Yellow
Model 5 All variables on 𝛽", PricePromotion on 𝛽# Orange Model 6 All variables on 𝛽"except PricePromotion Purple Model 7 ATL, BTL and PricePromotion on 𝛽" Black
The results of the analysis are shown in TABLE 9. Additionally figure 3 shows an example for the visualization of the different models of introduction 5. Overall, the analysis shows that models 6 and 7 have the highest model fit (Appendix 4). After the analysis, it has become clear that the models where the effect of the marketing variables are measured on 𝛽" (innovators) are better than on 𝛽# (imitators). For 𝛽", (baseline of external influence) and 𝛽$ (repeat) the same results are shown as the Petcare innovations, these two beta’s have a positively significant effect in almost every model. However, 𝛽# as well as the effect of the different marketing efforts are often negatively significant.
Table 9: Introductions Confectionary
Model Significant
Name 1 2 4 5 6 7 Positive Negative
Introduction 1
x
𝛽",, 𝛽$ 𝛽#, 𝛽".$ (BTL) Introduction 2x
𝛽",, 𝛽$ 𝛽# Introduction 3x
𝛽",, 𝛽$ 𝛽".#(EGKR),𝛽# Introduction 4x
𝛽",, 𝛽# , 𝛽$ Introduction 5x
x
𝛽",, 𝛽$ 𝛽# Introduction 6x
𝛽",, 𝛽$ 𝛽# Introduction 7x
𝛽$ 𝛽"."(display), 𝛽".#(EGKR), 𝛽".$ (EGDKW),𝛽# Introduction 8x
𝛽",, 𝛽$ 𝛽(EGDKW)#, 𝛽"." Introduction 9x
𝛽",, 𝛽$ 𝛽".#(EGKR),𝛽#To conclude, for confectionary innovations the most effective model is model 6. This model includes ATL, BTL, Display, EGKR, EGDKW on 𝛽" (innovators) (Appendix 4). In this model PricePromotion is left out, this is a variable which exists always and only when there is an in-store marketing effort. That is why there was a high correlation, which blurred the results. Thus, the original model which results from the literature, is not the most effective one. Putting al variables (with exclusion of PricePromotion) on 𝛽" (innovators) has the highest model fit.
Chapter 5: Discussion
As mentioned earlier there are different ways in which to influence customers to adopt an innovation. All kinds of marketing efforts have different effects on groups of adopters. Every company wants to be able to increase the initial purchases, called trail purchase and try to make an effort to bridge the gap between trail and repeat purchases to generate a successful innovation. After the analysis the perfect solution would be a universal model with a one size fits all. Unfortunately, but not surprising this is not what was found during this research. We found that there is a lot of heterogeneity in the models of the different innovations. Sometimes, marketing efforts are more effective on the external influence (𝛽"), meaning that the use of these instruments influences the purchasing behavior of innovators in buying an innovation. While, in other cases marketing is more effective on internal influence (𝛽#). This means that imitators are more influenced by marketing efforts in buying a new product. This heterogeneity also depends on the different type of customers you bind to you and how they react on new product introductions. Thus, marketing can play a part in how customers perceive the product or the speed with which they buy an innovation, but these efforts are not always decisive.
From the literature we learned that it pays off when a company will start investing heavily in media advertisement and in-store marketing efforts in the initial stages of the diffusion process to inform as many innovators as possible about the existence of the new product. This will result in innovators adopting the product, which in turn will provide information to the rest of the market: the imitators. When looking at the results of the analysis in this research paper, these findings were not completely consistent. The out-of-store marketing efforts theory and reasoning can be seen with the Petcare Dog innovations. Here the out-of-store marketing efforts and brand advertising was more effective on innovators than imitators. However, when looking at the results for the Petcare cat innovations and the Confectionary innovations the opposite is true. Here the out-of store marketing efforts and brand advertising was more effective on imitators than innovators. Furthermore, the in-store marketing efforts theory and reasoning can be seen with the Petcare Dog innovations as well as the Confectionary innovations. For these two segments, the in-store marketing efforts were more effective on innovators than imitators. However, the opposite has been found for the Petcare cat innovations. With these innovation the in-store marketing effort were more effective for imitators than innovators.
Third, direct product experiences have found to be effective in purchasing imitators to buy the innovation. In contrast to innovators, imitators want more information, take less risks, and want to experience the product before buying it. This line of reasoning is also confirmed within this research paper. In the analysis was found that sampling is more effective for imitators than innovators with Petcare innovations.
Further research on the best model fit found additional results. Striking is that when using the model with the highest fit for the Petcare innovations, it only includes the variables above-the-line and below-above-the-line advertising on external influence (innovators). Apparently these two marketing efforts best predict the sales of most Petcare innovations. Thus, the original model which results from the literature is not the most effective one for the Petcare innovations.
To add, when using the model with the highest fit for confectionary innovations it includes the variables ATL, BTL, Display, EGKR, EGDKW on external influence (innovators). The effect of the different marketing efforts are often negatively significant in the models with the best fit. While the literature shows promotions can generate short term gains, high promotion has a negative effect on the long run. This might explain the negative effect of these marketing efforts on the sales of the innovations. Eventually these marketing efforts will not generate more sales when they are measured on both external of internal influence. Moreover, external influence (𝛽") is often found to be negative significant as well. According to the literature, you would expect current users would positively influence the trail of future adopters. But apparently the experiences of current users do not always affect the purchasing behavior of new users. This might be explained by the fact that current users are not satisfied with the innovation, due to quality issues or if the innovations does not sufficiently meet the customer needs.
Furthermore, for both categories looking at the highest model fit as well as the models used in the hypothesis analysis, the external influence (𝛽",) and repeat purchases (𝛽$) have the highest positive influence on sales, meaning these parts of the diffusion process add the most to the amount of sales generated of the innovation.
