‘’TRADITIONAL VERSUS NEW MARKETING
INSTRUMENTS - WHAT DO CONSUMERS
VALUE MOST?’’
‘’TRADITIONAL VERSUS NEW MARKETING
INSTRUMENTS - WHAT DO CONSUMERS
VALUE MOST?’’
F.F. (Fernande) Brand
University of Groningen
Faculty of Economics and Business
MSc Marketing Research and Management
MANAGEMENT SUMMARY
This research paper examines the effects of traditional marketing instruments versus new marketing instruments on consumers' online purchase intentions of fast moving consumer goods. The traditional marketing instruments researched in this study were advertising, pricing and branding. New marketing instruments researched on the other hand were online consumer reviews and third party reviews. 292 respondents participated in an online choice-based conjoint analysis to capture the relative importance of each marketing instrument on the purchase intention, and to get under the skin of what consumers really value in making a purchase decision of fast moving consumer goods online.
Research findings especially provide new perspectives and insights to managers of firms operating in the FMCG industry, on how to manage OCR and third party reviews in comparison to more established marketing instruments like advertising, pricing and branding on their web sites as being part of their online marketing strategies.
This study investigates that price is the most important dimension in determining the purchase intentions of FMCG online, followed by third party reviews, branding and OCR. The positive effects of high third party reviews and high OCR on purchase intentions are stronger compared to negative effects of low third party reviews and low OCR. Although firms operating in the FMCG industry are the biggest online ad spenders, results of this study show that advertising is the least valued attribute.
Furthermore, latent class analysis showed different effects for the marketing instruments between different classes regarding purchase intentions. Meaning marketing managers in the FMCG industry should carefully consider their customers while developing an optimal online marketing strategy.
Results showed that class one could be best targeted by positive online review results, both from consumers and third parties. Consumers in class two, four and five can be best targeted by specific brand offers, since in these classes are strong specific brand preferences. For targeting consumers in class three, the pricing strategies should be taken stronger into account. Because in class one and class three are no specific brands preferred, in these classes price and reviews are the decisive factor in consumers’ purchase intentions.
The presented model is expected to help firms in the FMCG industry in the future to design relevant online marketing strategies and delivers insights in interesting fields for further research on where and how to spend their online marketing budgets.
PREFACE
In 2008 I started studying at the University of Groningen. Within three years, I accomplished my Bachelor degree in Business Administration. After that, I decided to do my Masters in the field of Marketing. This was an easy choice for me, because since I started studying, most of my interests were in the field of marketing. I decided to follow both tracks, Marketing Management and Marketing Research, which I both enjoyed a lot. Especially the high quality of the Master courses was something I highly valued, because the drive of the professors gave me extra motivation. I really want to thank all the professors and other employees who are working in the Marketing department for the great effort they put in teaching and setting up the courses, it really makes a difference.
In my Master year, I decided to do some extracurricular activities, something I can recommend to all students. In 2011-2012, I participated in the International Business Research (IBR) project Indonesia, which was truly the best choice I made in my whole study period. New friendships, practical experience in doing market research, being in contact with companies, travelling, I enjoyed every moment of this project. After that, I went to Paris for half a year, for a Marketing and Business development internship at a fun and growing travel company called Crazy-Voyages. Although it sometimes was not so easy to live there all by myself, I learned a lot from this experience. Last year I got the chance to further develop my practical experiences at Heineken as a Marketing Intern. At Heineken I could bring what I had learned into practice and really got to see ‘’the world of Marketing’’.
All these experiences made my study period very complete.
And now, it is time to say goodbye to this beautiful period.
This Master Thesis represents the final phase of my study. I want to thank my supervisor dr. H. (Hans) Risselada for his time, his guidance, and the useful insights he gave me during the writing of this Thesis.
1
Table of Contents
MANAGEMENT SUMMARY
PREFACE
1. INTRODUCTION
3
2. THEORY
6
2.1 Purchase Intention of FMCG
6
2.2 Online Consumer Reviews (OCR)
6
2.3 Third Party Reviews
8
2.4 Online Advertisements
10
2.5 Price
11
2.6 Brand
11
2.7 Relative Importance’s of Attributes
12
2.8 Moderating Role of Brand Importance
13
2.9 Consumer Descriptive
15
2.10 Conceptual Model
16
3. RESEARCH DESIGN
17
3.1 Methodology
17
3.2 Model Specification and Estimation
18
3.3 Stimuli
18
3.4 Design
20
3.5 Data Collection
22
3.6 Control Variables
22
3.7 Segment Solution
23
3.8 Covariates Related to Class Membership
23
3.9 Predictive Validity
23
4. RESULTS
24
4.1 Descriptive Statistics
24
2
4.1.2 Online Behavior
25
4.1.3 Beer Consumption Behavior
25
4.1.4 Brand Importance
25
4.2 Choice-Based Conjoint Analysis - Aggregated Model
26
4.3 Interpretation of the Hypotheses on Aggregated Level
28
4.4 Predictive Validity
28
4.5 Choice-Based Conjoint Analysis - Segments
29
4.6 Latent Class Model: Interpretation of the Segments
32
Class 1: The Online Review Evaluators
32
Class 2: The Heineken Brand Lovers
33
Class 3: The Price Conscious
33
Class 4: The Hertog Jan Brand Lovers
34
Class 5: The Grolsch Brand Lovers
34
4.7 Moderating Effects of Brand Importance
34
4.8 Overview of the Hypotheses
36
4.9 Predictive Validity
37
5. DISCUSSION
38
5.1 Implications
40
6. LIMITATIONS AND FUTURE RESEARCH
43
7. LITERATURE
45
8. APPENDICES
50
Appendix I - Entire Questionnaire
50
3
1. INTRODUCTION
The advent of the internet has allowed consumers to find a lot of information online (Maeyer, 2012). Where consumers in the past always relied on information from relatives and acquaintances to find out about product attributes and quality (word-of-mouth), nowadays, a new type of WOM information, called electronic-WOM (eWOM) is rapidly developing. Results of many studies in this field show that online reviews, as an important part of eWOM, is an emerging market phenomenon that is playing an increasingly important role in consumers’ purchase decisions (Chen and Xie, 2008; Feng and Xiaguan, 2010; Meayer, 2012). Therefore it has to be taken into account by managers of firms.
Online reviews are an important part of eWOM. Nowadays, a lot of firms already use it as a new important online marketing tool (Feng and Xiaguan, 2010). Ante (2009), mentions that whereas online reviews first became popular for specialty products, nowadays it is common for consumers to look up reviews even for relative commodities such as cleaning products (FMCG). Also, Chen and Xie (2008) note in their research that online reviews are common for many product categories such as books, electronics, games, videos, music, beverages and wine, of which beverages and wine can be seen as being part of the FMCG industry.
Therefore, we expect that these relatively new marketing instruments also have an influence on consumers' purchase intentions for FMCG and that online reviews can be used as a new marketing tool in the FMCG industry as well.
According to Trusov, Bucklin and Pauwels (2009) WOM marketing can be even more effective than traditional marketing activities like advertising. A study of Cho and Cheon (2004) also shows negative trends in internet advertising, such as "banner blindness" and extremely low click-through rates. Many study findings in this field of research show that traditional forms of communication appears to be losing their effectiveness (Prendergast, Ko, and Siu Yin, 2010; Podnar and Javernik, 2012; Tseng and Chen, 2013) Thereby, Ho-Dac, Carson, and Moore (2013) stated in their research paper that easy access to online consumer reviews (OCRs) has led some observers to posit that brand names, as assurances of product quality and performance, will lose much of their importance in the interactive marketing environment. This line of reasoning suggests that customers will bypass marketer-influenced signals such as brands and instead rely directly on unfiltered word of mouth (like online reviews) from other consumers or third parties.
In the FMCG industry, brands are viewed as the key assets of the company, and all investments will be made to create strong brands (Schuiling, 2004). Also, according to the Internet Advertising Bureau (IAB) and PwC’s advertising expenditure report (2012) FMCG’s share of the online advertising market has almost doubled in the last three years. Firms operating in FMCG are nowadays amongst the biggest online ad spenders (O’Reilly, 2012).
4 marketing budgets in the right way, or do they spend too much money on traditional marketing instruments like online ads, pricing and branding?
Firms are nowadays under increasing pressure to justify their marketing expenditures. This evolution toward greater accountability is reinforced in harsh economic times when marketing budgets are among the first to be reconsidered (Heerde, Gijsenberg, Kimpe and Steenkamp, 2013). To make appropriate decisions about where to spend marketing budgets, managers must know whether, and to what extent, marketing instruments effectiveness varies; however, surprisingly little research addresses this issue in the FMCG industry.
In this research paper we will therefore take the effects of different marketing instruments into account to be able to measure which marketing instruments are the most important in the purchase decisions of consumers. We will take traditional instruments like advertising, pricing and branding, together with more recently developed marketing instruments like online consumer reviews and third party reviews into account.
In this research online reviews will be separated in two different types, namely online consumer reviews (OCR) and online third party reviews. The two types of online reviews can both have a different effect on purchase intentions (Maeyer, 2012). Research of Godes and Mayzlin (2009) pointed out that third party reviews, for example reviews from opinion leaders or consumer associations are more effective than online reviews of regular consumers. Both types of online reviews will therefore be taken into account in this research paper.
There exists the view that firms need to actively manage eWOM, given the great efficiency of the internet in spreading WOM, and that they should also strategically respond to the existence of online reviews (Chen and Xie 2005; Dellarocas 2006). This study adds insights to this view especially for firms operating in the FMCG industry by answering the following main research question.
How do traditional versus new marketing instruments affect consumers’ online purchase intentions of fast moving consumer goods?
5 An online choice based conjoint analysis has been conducted among 292 Dutch respondents in order to expose people’s preferences and investigate the contribution of each marketing instrument on consumers' purchase intentions for FMCG.
This research expands on previous findings in several ways. First, the importance of online reviews has been discussed in many research papers (Maeyer, 2012). It is clear that many firms already use online reviews as a marketing tool. FMCG, as a product category gains disproportionate amounts of attention in research about these relatively new marketing instruments, compared to other product categories. Also, the effects of online reviews (OCR and third party reviews) on consumers’ purchase intention on FMCG, compared to effects of more traditional marketing instruments like advertising, pricing and branding have not been studied in a conjoint setting before.
Insights in which attributes are most valued by customers is also very relevant for practitioners in this field. With a shift from offline shopping to online shopping these are important issues for organizations operating online. Firms operating in the FMCG industry can use the insights of this research as input for decision making in the future about incorporation of, – and how to incorporate these new marketing instruments in their online strategies, to increase consumers' purchase intentions and thereby the company’s success.
6
2. THEORY
This chapter will give an overview of existing literature about the variables taken into account in this research paper. First, we will start by describing our dependent variable - purchase intention. Then we will briefly describe the independent variables (attributes) one by one. The independent variables taken into account are online consumer reviews (OCR), third party reviews, advertising, price and brand. Hypotheses are formed based on the existing literature about these marketing instruments. The chapter will conclude with a conceptual model.
2.1 Purchase Intention of FMCG
Purchase intention is the dependent variable in this study. This means that we measure the effects of the independent variables on consumers' purchase intentions of FMCG. We define purchase intention as consumers’ buying behavior and decision makings (Soonyoung and Taesik, 2010). In other words, purchase intention measures consumers’ intentions (willingness) to buy certain products and thereby it is a sales indicator. Weaker purchase intention leads to lower purchases, whereas stronger purchase intentions lead to higher purchases of customers (de Canniére, de Pelsmacker and Geuens, 2009). Fast moving consumer goods (FMCG) is the industry wherein we execute this research. FMCG have been defined as frequently purchased essential or non-essential goods such as food, toiletries, soft drinks, disposable diapers etc. (Business dictionary). Examples of typical global operating firms in this industry are Coca-Cola, Nestlé, Heineken or Kraft (Schuiling, 2004). Online shopping of fast moving consumer goods (online grocery shopping) has not been accepted as fully as other types of online purchasing, such as books, electronics, or products in the travelling branch (Freeman and Freeman, 2011). However, these amounts are growing (Junhong et al., 2010).
2.2 Online Consumer Reviews (OCR)
The first independent variable in our model is OCR. We will define OCR according to Chen and Xie (2008) as a type of product information created by users based on personal usage experience, that can serve as a new element in the marketing communications mix that can work as free “sales assistants” to help consumers identify the products that best match their idiosyncratic usage conditions.
7 There are several different forms to show OCR, for example in the form of ratings, or more extended forms with much more text elements. Floh, Koller and Zauner (2013) found that people that are concerned with low priced goods (the product category in which FMCG can be placed as well), read reviews in less detail than people that are more concerned with purchasing high priced goods. Meaning that consumers who want to buy low priced goods may focus more on review ratings than on detailed text messages.
In this research the focus will therefore be on online reviews in the form of numerical star ratings. Online ratings are a quantitative summary of experiences, attitudes, and opinions, usually expressed as stars (Floh, Koller and Zauner, 2013). Although more extended reviews and ratings are often used in the same context, these concepts are different. A review provides a qualitative assessment of one’s product experience, while a rating is a rather quantitative evaluation (Sridhar & Srinisivan, 2012). In numerous leading web stores, like Amazon.com, Bol.com, and Apple’s iTunes store, the average rating is prominently displayed to convey information about product evaluations. Star ratings for online customer reviews typically range from one to five stars. A very low rating (one star) indicates an extremely negative view of the product, a very high rating (five stars) reflects an extremely positive view of the product (Mudambi and Schuff, 2010).
It is commonly understood that online consumer reviews can reduce consumer uncertainty about product characteristics and, therefore, have the potential to increase product demand and firm profits (Maeyer, 2012). Recent evidence of the research of Chen and Xie (2008) suggests that OCR have become very important for consumer purchase decisions and product sales. Regardless of whether they are positive or negative, eWOM communications have been shown to directly influence consumer attitudes and behaviors (Jones, Aiken and Boush, 2009). For example, Chevalier and Mayzlin (2006) find that online consumer ratings significantly influence product sales in the book market and Zhang and Dellarocas (2006) obtain similar results in the movie industry. Positive online consumer reviews are indicative of a product’s quality and reputation and increase product sales. Negative reviews signal a lack of confidence in the product among consumers and decrease product sales (Geng, Hon-Kwong and Xiaoning, 2012). Soonyong and Taesik (2010) also studied the effect of OCR on product sales and concluded that both positive and negative OCR have a significant influence on product sales.Although recent research about the influence of OCR on sales has shown that the star rating of a review significantly affects sales (Chevalier and Mayzlin, 2006; East, Hammond, and Lomax, 2008 and Liu, 2006), this has not been explicitly researched for products in the FMCG yet. To be able to confirm this for the FMCG industry, we developed our first hypothesis which is in line with results from prior studies in other industries.
H1a OCR have a positive effect on purchase intentions of FMCG
8 Chevalier and Mayzlin (2006) also found that product sales are influenced by negative reviews more than by positive ones. It implies that consumers’ perception of an online consumer review can be different depending on review ratings. Park and Lee (2012) argued that consumers tend to think that negative electronic word-of-mouth is more credible.
A widely accepted explanation for the impact of negative WOM is the so-called negativity bias, a psychological tendency for people to give greater diagnostic weight to negative information in making evaluations (Geng, Hon-Kwong and Xiaoning, 2012). According to these prior research findings, we hypothesize the following:
H1b Negative star ratings have a stronger effect on purchase intentions of FMCG than positive star ratings have
Although research already has shown the influence of positive and negative OCR on purchase intention, this hypothesis still adds interesting insights. OCR has not been researched in a conjoint setting together with more traditional marketing instruments like advertising, pricing and branding. Also, this concept gains disproportionate amounts of attention in the FMCG industry, while online development in this sector move on. It is very interesting to see how big the influence of star ratings is compared to other, more established marketing instruments in the FMCG industry. It is also interesting whether or not for this industry can be confirmed that negative OCRs are more influential than positive OCRs are, like prior different research results has shown.
2.3 Third Party Reviews
Third party product reviews based on independent laboratory tests, product comparisons and recommendations or expert evaluations have grown increasingly popular in recent years. Various popular consumer magazines (e.g., PC Magazine, PC World, Consumer Reports, Car and Driver, Scuba Diving Magazine, Runner’s World, Entertainment Weekly, Gourmet) regularly publish comprehensive reviews of products of interest to their readers. Moreover, the internet and fast-developing information technology have significantly reduced reviewers’ information-delivery cost and consumers’ information-retrieval cost. As a result, a growing number of websites (e.g., CNET.com, ZDNET.com, caranddriver.com, swiminfo.com, wireless design. com, enjoythemusic.com, golfdigest.com) are offering online third-party product reviews. In addition, consumers can now easily access and compare product reviews (Chen and Xie, 2005).
9 consumers than from third parties.
Online sellers can invite users of their products to post personal product evaluations on the sellers’ websites, but in addition to that they can also provide their customers with review information offered by third party sources. An effective use of online third party reviews help consumers reduce their perceived risk of online shopping and stimulate them to buy online (Soonyong and Taesik, 2010). According to Senecal and Nantel (2004), it is because of consumers’ perception that other parties have more useful information about the product than themselves. This is true for both the offline and online shopping environment. In particular, product recommendations from third party reviews can help consumers reduce their search cost and stimulate them to buy online (Senecal and Nantel, 2004). Therefore, the following independent variable that we take into account in our model is online third party reviews. In this research paper we investigate the option for a marketer of adding a third-party feature to their company website, as an addition on online consumer reviews, pricing, advertising and branding. Where, obviously, the third party – the creator of it, is identified.
Market observations suggest that third-party product reviews have a significant effect on the success/failure of products. For example, USA Today reported that “a bad review in a computer magazine can kill a product and often does’’ (Chen and Xie, 2005). A survey by the Wall Street Journal showed that over a third of Americans sought the advice of critics when choosing a movie (Simmons, 1994). And in a survey reported by The Los Angeles Times, 44% of online consumers said they consulted independent review websites before making a purchase (Chen and Xie, 2005). A research about craft beers showed a significant influence of online third party reviews on purchase intentions of craft beers. In the market for craft beers, retailers report significant improvements in the ability to market highly differentiated and high-priced beers based on reviews provided by RateBeer (www.ratebeer.com), a high-traffic specialty review site dedicated to this market (Li, Hitt and Zhang, 2011). Beer is a good example of FMCG. However, craft beers are seen as niche products and not as common mass-market products, like common craters of well-known beer brands that you will find in a regular supermarket. Still, this finding gives an indication that online third party reviews are expected to have an influence on FMCG. Godes and Mayzlin (2009) found that third party reviews, for example reviews from opinion leaders or consumer associations are more effective than OCR.
While these recent studies show evidence for the influence of third party product reviews on purchase intentions, researching this variable still makes an important contribution, since this is an under-explored area especially in the FMCG industry. We therefore want to measure if this third party reviews have a positive influence on purchase intentions in the FMCG industry and if they indeed have a stronger influence than online consumer reviews have. To test this relationship we developed the following hypotheses:
H2a Third party reviews have a positive effect on purchase intentions of FMCG
10
2.4 Online Advertisements
The quantity of online media, advertisers, and advertising space is increasing, thereby augmenting the amount of information available to consumers. The modern media are so overwhelmed with advertising messages that consumers are not able to process all of them, which means that advertising is becoming less effective (Podnar and Javernik, 2012). Also, Tseng and Chen (2013) state that traditional forms of communication, such as advertising, appear to be losing their effectiveness, possibly because consumers doubt their reliability and trustworthiness (Prendergast, Ko, and Siu Yin, 2010). A study of Cho and Cheon (2004) also shows negative trends in internet advertising, such as "banner blindness" and extremely low click-through rates. According to these findings, online advertising seems to lose credibility. Since firms in the FMCG industry are the biggest spenders of online ads, it is interesting to study if online ads really show high effects on customers’ purchase intentions. What role does online advertising play compared to new marketing tools like OCR, third party reviews, and should firms operating in the FMCG industry also consider investing money in these fields?
The internet and information technology represent a new opportunity for consumers to share their product evaluations online. The increasing amount of information and media fragmentation are making it more difficult for traditional marketing tools, such as advertising to reach their target audiences. The same cannot be said for online word of mouth. The credibility of WOM, combined with the fact that consumers are more involved, suggests that WOM has increasing effects compared to advertising (Brown, Broderick and Lee, 2007).
On the other hand, Spaulding (2010) argues that it is effective to advertise online, since ads can produce results as long as the participants do not mentally or technically block them. Purchase intentions of FMCG can be easily changed since it is a low risk purchase (Schuiling, 2004) and therefore switching costs are very low. So advertisements can be effective in turning people’s choice for another product, and thereby increasing sales. Looking at this reasoning, online ads should be effective for FMCG.
Research of Tseng et al. (2013) however shows that ads have no significant positive effect on consumers’ purchase intention. However, this was specifically researched within consumers that are member of virtual communities of products like mini-cooper, and not for FMCG in specific. Since it was in a virtual community, the members relied mostly on opinions from community members than they did on advertising. In virtual communities, products become popular through WOM and require little marketing in these communities (Tseng et al., 2013). We expect a different effect of online ads on purchase intentions of FMCG. FMCG are much cheaper products (than cars). Switching cost for consumers are low (Li, Hitt and Zhang, 2011) and therefore we expect that ads can account for switching from one product to another. Also, this research is outside of a virtual community, where the consumers focus not only on preferences and opinions of others (members of the community). We therefore expect an effect of online ads on purchase intentions in the FMCG industry and developed H3 as follows.
H3 Online advertisements have a positive effect on purchase intentions of FMCG
11 very relevant for firms operating in these industries to know the effects of online ads in their businesses.
2.5 Price
For the relevance of the conjoint study, it is important to include price as a variable influencing our dependent variable - purchase intention. It is generally known that price is amongst the most important influencers on purchase intention of products and services in general, this also counts for FMCG. Whether a consumer is buying online or offline, price is in a lot of cases the number one decision influencer in the decision making process for a consumer when buying a product (Kuan-Pin, Chiang and Dholakia, 2003). A price can be defined as: “the consumer's perceptual representation or subjective perception of the objective price of the product” (Kuan-Pin, Chiang and Dholakia, 2003).
Dutch consumers are basing their online purchase decisions mostly on price; research of Intomart GfK (2013) showed that 89% of Dutch consumers based their decision on the decisive factor price. Also in this research it is expected that price plays a major role in purchase intention of people.
Bijmolt, Van Heerde, and Pieters (2005) report an average price elasticity of -2.62, meaning that higher prices in general lead to lower purchase intentions. They also have shown that over time, consumers have even become increasingly price sensitive. We therefore hypothesize the following.
H4 Price has a negative effect on purchase intentions of FMCG
2.6 Brand
We define brand as a name (or symbol) that is given to a product that will differentiate it from other products and that will register it in the minds of consumers as a set of tangible (rational) and intangible (irrational) benefits (Higie and Sewall, 1991). A brand can be considered as the added value of investment made in a product. Brands create differentiation which lead to brand preferences. Especially in the FMCG industry brands are very important and firms invest a lot of money in brand development (Schuiling, 2004). Brand will therefore be included in this study as our last independent variable. In this study we take a look at whether or not consumers have a preference for a brand and what the influence on consumers' purchase intentions of the attribute brand is. Hellier et al. (2003) defined ‘brand preference’ as ‘the extent to which the customer favors the designated service provided by a certain company, in comparison to the designated service provided by other companies in his or her consideration set’. Given that customers are often loyal to multiple brands, they form brand preferences for more than one brand in most purchase situations (i.e. polygamous rather than monogamous), although they ultimately have to choose one for purchase at a time (Kim, Ok and Canter, 2012). According to Higie and Sewall (1991), brand preference is a major component of individuals' expressed intentions to purchase a particular brand and we therefore hypothesize the following.
12
2.7 Relative Importance’s of Attributes
By studying the influence of these five attributes in a conjoint setting, we are able to find out what factors have the highest effect on consumers’ purchase intention. Prior research already showed findings that give indications about importance of the different attributes. We therefore developed some expectations about the relative importance of our attributes.
For example, existing empirical generalizations indicate that price elasticities are typically 10-20 times larger than advertising elasticities (Bijmolt, Van Heerde, and Pieters 2005; Sethuraman, Tellis, and Briesch 2011). This supports the argument that pricing is of high importance in determining purchase intentions. However, will price still be the most important variable affecting purchase intentions of FMCG? Or do consumers value the other attributes more? There is as far as we know no prior research to the influence of OCR and third party reviews on purchase intention relative to pricing, advertising and branding. This, together with that it is generally known that price is amongst the most important influencers of purchase intention of products and services in general, we expect that price in this research is also of the highest importance. Thereby, as already mentioned, research of Intomart GfK (2013) showed that Dutch consumers base their online purchase decisions mostly on price. Bijmolt et al., (2005) report an average price elasticity of -2.62 and also have shown that over time, consumers have become increasingly price sensitive.
We therefore expect that price is the strongest influencer amongst all our independent variables. This results in the following hypothesis:
H6a Price has stronger effects on purchase intentions of FMCG than OCR, third party reviews, advertising and branding have
Although advertising always seemed an important marketing weapon, research about advertising elasticities from the last 25 years showed that short-term advertising elasticities for established products are very small (approximately .01) (Heerde, Gijsenberg, Kimpe and Steenkamp, 2013). In a recent meta-analysis, Sethuraman, Tellis, and Briesch (2011) find a mean long-term elasticity of .24 across 402 observations, with 40% of these elasticities between 0 and .1 which is also low. Sethuraman et al., (2011) also find that advertising elasticity is lower in more recent studies, which suggests that advertising elasticity declines over time. The last period (1984–2008), that was researched by Sethuraman et al., (2011) has witnessed significant changes on many fronts that may have an impact on the effectiveness of advertising. The marketing environment has changed as a result of greater competition, globalization and the advent of the Internet (Heerde et al., 2013).
13 one of the most effective marketing tools (Jiménez and Mendoza, 2013). Traditionally consumers exchanged WOM through face-to-face conversations. However, as consumers increasingly use the internet to communicate with other consumers as well as to review and purchase products, OCR have become an important form of WOM (Trusov et al., 2009). Because of the importance of WOM, that nowadays transforms to eWOM, together with the decreasing importance of advertising, we believe OCR and third party reviews (as part of eWOM) to be more important in consumers’ purchase intentions compared to the importance of advertising.
Branding has become one of the most important determinants of consumer choices (Philiastides and Ratcliff, 2013). Researchers agree that the choice of brand name for a product can alter the consumers' judgment about the product and their purchase intention process. With competition getting fiercer and product quality becoming more homogenous, a “better” brand name can be decisive in product choice if the consumers compare several products.According to Higie and Sewall (1991), brand preference is a major component of individuals' expressed intentions to purchase a particular brand. Advertising is a way you can do brand expenditures, to in the end, build the brand and create stronger brands (Fischer, Volckner and Sattler, 2010). We could say brand is already on a higher level. When there is a preference, people would have less eye for advertisements. Therefore we also expect the effect of brand to be higher than the effect of advertising.
Although research has shown that the effectiveness of online ads decreases nowadays, we expect online ads in FMCG industries to still have a positive effect (H3) on consumers’ purchase intentions, however we expect, because of the above standing arguments, the effect of the other attributes to be stronger. Therefore H6b is developed as follows:
H6b OCR, third party reviews, pricing and branding have a stronger effect on purchase intentions of FMCG than advertising has
2.8 Moderating Role of Brand Importance
14 of brand equity from an information economics perspective. The main reason people make use of online reviews (OCR and third party reviews) is also risk reduction (Feng and Xiaguan, 2010). Brands identify the source or maker of a product. Consumers recognize a brand and activate their knowledge about it (Fischer et al., 2010). Using what they know about the brand in terms of overall quality and specific characteristics, consumers can form reasonable expectations about the functional and other benefits of the brand. Consequently, brands contribute to reducing the consumer’s (subjective) risk of making a purchase mistake (Fischer et al., 2010). According to this, we could also say that if consumers find brand very important, they use their brand preference as a signal to reduce perceived risk. They would then not feel the need for (or have less need for) other risk reducers like OCR and third party reviews. In the end this would mean that if brand importance is high, this would decrease the effect of online consumer reviews or third party reviews on purchase intentions. We therefore hypothesize the following:
H7a Stronger Brand importance decreases the positive effect of online consumer reviews (OCR) on purchase intentions of FMCG
H7b Stronger Brand importance decreases the positive effect of third party reviews on purchase intentions of FMCG
Higher brand importance indicates that brands are more relevant to the customer. As a consequence, customers are more inclined to rebuy the same preferred brand and are willing to pay more for that brand (Fischer et al., 2010). Chakravarti and Janiszewski (2004) also say that if brand importance increases, the substitutability of one brand for another decreases and price elasticity decreases as well. This means that the effect of price decreases, which means that brand importance moderates the effect price has on consumers’ purchase intention. In line with this we hypothesize that:
H7c Stronger Brand importance decreases the negative effect of price on purchase intentions of FMCG
In markets in which customers are more brand sensitive, demand is also more responsive to brand expenditures. Assuming profit maximization, the Dorfman–Steiner theorem (Dorfman and Steiner, 1954) recommends guiding larger brand resources to these markets. Advertising is one way in which you can do brand expenditures. In a model where brand importance is high, these advertisements will be more effective, since, when the role of brands is important for consumers, consumers will be more aware of advertisements and will get more involved with brands (Miller and Berry, 1998). Therefore the attitude of consumers to this attribute will increase because if consumers value the importance of brands they are more open to communications of brands and as a consequence the effectiveness of advertisements will increase. We will therefore hypothesize the following
15 Specifically, when customers believe that brands are important for their buying decision, they do so because brands provide important functions along the purchase decision and consumption process (Fischer et al., 2010). According to Higie and Sewall (1991) brand preference is a major component of individuals' expressed intentions to purchase a particular brand. Fischer et al. (2010) also found that high brand importance accounts for higher commitment to rebuy a preferred brand in the future, thus causing repetitive same-brand purchasing. When there is higher brand importance, customers have a greater demand for brand benefits, such as reduced risk, and the brand name plays a pronounced role in the buying decision. In line with these findings we expect that brand importance moderates the effect brand has on purchase intention. The more important consumers find the role of the brand, the higher the effect of our independent variable brand on consumers' purchase intention. Our following hypothesis is therefore constructed as follows:
H7e Stronger Brand importance increases the positive effect of brand on purchase intention of FMCG
The hypothesized moderating effects of brand importance could be interesting for both managerial and academic purposes. As been stated before, firms operating in the FMCG industry invest a lot of money in developing brands. If brand importance really moderate the effects of the other attributes on purchase intention, this shows even more how important branding still is for organizations. Also, this is interesting with an eye on new marketing instruments like online consumer reviews and third party reviews. If brand importance strongly decreases the effect of these variables on purchase intention this means it is worth investing in a brand, because it can protect a product from negative information in the form of online reviews.
2.9 Consumer Descriptive
16 findings have practical implications for online sellers to guide them to effectively use online consumer reviews to engage females in online shopping. All these consumer differences are very important to take into account in this research since we will try to develop different consumer segments. By taking consumer differences into account, we are better able to describe different consumer segments later on in the conjoint analysis.
2.10 Conceptual Model
The expected effects are presented in the conceptual model as visible in figure 1. We expect that OCR, third party reviews, and advertising all have a positive influence on our dependent variable purchase intention. Expected is that brand has a strong effect on purchase intentions and that price has a negative influence on purchase intention. Furthermore, the moderating role of consumers’ value on brand importance is expected to be negative for OCR, third party reviews and price, and positive for the attributes advertising and branding.
H6a H6b - + + + + OCR H1a H1b
Third party reviews H2a H2b
Advertising H3
Brand H5
Purchase intention
Brand importance H7a-e
Figure 1: Conceptual Model
Descriptive
17
3. RESEARCH DESIGN
This chapter presents the research design. The data needed for this research is collected by doing quantitative research. In this chapter first the methods used in order to find empirical evidence to support the above expected relationships will be described. After that we will describe in detail how the survey was set up, how data in this study was collected and how was dealt with the data thereafter.
3.1 Methodology
Conjoint Analysis is a multivariate statistical technique developed to understand how respondents develop preferences for any type of product or service based on the premise that consumers evaluate the value (utility) of a product or service by combining the separate amounts of value provided by each feature (Hair, Black, Babin and Anderson, 2010). The features are called attributes in conjoint analysis. The attributes get different levels, by measuring these levels by conducting Conjoint Analysis, you can get ‘’under the skin’’ of what attributes people value the most. It determines the relative importance consumers allocate to a set of attributes, and the utilities consumers attach to the level of attributes. To study consumer preference based on utilities, products are considered as scores on a set of attributes; the utility of a product equals the sum of the utilities of the preferred attribute levels (Wierenga, 2008). To be successful in defining utility, the object in terms of its attributes and all relevant values for each attribute has to be described (Hair et al., 2010). The different choice sets (in this case nine sets of three choices) of potential products or services is shown to respondents. By analyzing how they make preferences between these products, the implicit valuation of the individual elements making up the product or service can be determined. Conjoint analysis can also isolate groups of potential customers who place differing importance on the feature to define high and low potential segments. Especially this is valuable information for marketing managers, because with this information, market segments can be targeted with favorable conditions, when specific preferences are known. We therefore also develop a segmented model in this research.
18
3.2 Model Specification and Estimation
Since the utility of a purchase intention for products in the FMCG industry (in this case used as product: beer craters) equals the sum of the utilities of the attribute levels, the formula to determine utility (U) per segment (j) is:
Where, U= Utility
j= segment 1, …, n X₁ = OCR
X₂ = Third Party Review X₃ = Advertsing X₄ = Price X₅ = Brand
To get an ‘’ideal’’ utility we have to take a look at the parameters (utilities) of the preferred levels per attribute. This ideal utility is different for every class. In the interpretation part of the segments in section 4.6 the preferred levels per attribute will be discussed.
The dependent variable in a CBC analysis is the choice - which item is selected from a set of alternatives - a respondent makes. The choice a respondent makes is assumed to be the alternative a consumer finds the most useful out of the available alternatives. The attributes (OCR, third party reviews, advertising price and brand) are the explanatory variables. The relative importance of each attribute can be calculated by dividing the largest part-worth difference within an attribute by the sum of the largest differences of all attributes (Hair et al., 2010).
3.3 Stimuli
In this research five different attributes are included, each with three attribute-levels. The five attributes are OCR, third party reviews, advertising, price and brand. As mentioned in the theoretical framework, these are found to be attributes that are expected to have a certain influence on consumers’ online purchase intentions in the FMCG industry. Table 1 gives an overview of the attributes and their associated levels.
Attribute Level 1 Level 2 Level 3
OCR 4.5 stars
Third Party Review Position 8 out of 10 Position 1 out of 10
Advertising Customized Advertising
Price €9.99
Brand Grolsch
19 The product chosen as example for fast moving consumer good is a crater of beer, it is a typical fast moving consumer good and a very common product, that nearly all consumers buy, or once have bought. This study measures online purchase intentions of FMCG, and since not all respondents are frequent online grocery (FMCG) shoppers, beer is a relevant product for this study. Beer is one of the products people are very open for to buy online. It is heavy to transport a crater of beer, therefore it is an advantage for consumers to buy online, because delivery of crater(s) of beer is very helpful for them (Heineken, Q2, 2013). This makes it easy for the consumer to imagine them buying a crater of beer online. In our questionnaire we included some images of an online grocery shopping environment so that consumers can easily imagine that they are doing an online purchase. In this way the research setting is more comfortable. This makes it easier for the respondent to fill in the questionnaire.
Furthermore, the attribute levels are based on what is common in the Netherlands. The first attribute; online consumer reviews (OCR) has three levels, 1.5 stars, 3 stars and 4.5 stars. It was decided to not take the most extreme ratings like zero stars and five stars, since research showed that extreme values can be seen as less realistic and thereby are less trusted by consumers (Maeyer, 2012). On the other hand, it was important to take levels where the difference between the levels was big enough. Therefore it was decided to take those three levels No extremes, but they show clear differences between them.
The third party reviews vary between position no 1 out of 10 with grade 9.5, position 4 out of 10 with grade 6.5 and position 8 out of 10 with grade 3.5. Dutch people are used to give grades on a scale from 1 – 10 so this is common for them. Position 1 out of 10, 4 out of 10 and 8 out of 10 means on which position the beer brand is rated by the third party (amongst ten brands). We choose these levels because we want to see the different effect of a very positive third party review (a recommended product) with rating 1 out of 10, an average rating (level 4 out of 10) and a low rating (level 8 out of 10). We choose out of 10 brands because the third party consumer TV-program Kassa recently did an independent beer test, where experts blindly tested ten different beer brands on their taste. Heineken became no. 1 out of the 10 brands according to the experts. As third party therefore Kassa was taken. Kassa is a well-known consumer TV-program from the VARA which deals with consumer complaints, tips, tests, and the latest news. In this research the impact of those third party reviews (like the reviews from Kassa) on consumers’ purchase intentions will be measured, when pro-actively shown as a marketing tool in an online grocery shopping environment.
For advertising three levels are being used as well; no advertising, general advertising and customized advertising. With customized advertising we do not mean adapted to the customer, but we mean the advertisement is adapted to the brand. In other words; the advertisement is specific for the brand. The most recent advertising campaigns of the different brands are used as the customized advertisements. As a general advertisement the word ‘’beer’’ is used, something you also see in retail advertising folders of different supermarket formats.
20 discounts on craters of beer. The price differences used here are very realistic compared to the current prices in the beer market.
The last attribute-levels are the brands; the brands chosen for this research are Heineken, Hertog Jan and Grolsch. All three are premium Brands in the Netherlands and all three are well known by consumers. They are also all independent (they do not fall in the product portfolio of one single company, like for example Amstel, Brand and Heineken, these brands all fall under the product portfolio of Heineken). However, consumers can have a preference for one (or more) of them. After deciding on the attributes and attribute-levels, the profile presentation has to be chosen. In CBC analysis three methods of presentation are associated: pair-wise comparison presentation, trade-off presentation and full-profile method. For this study, a full-profile method is chosen because it is most appropriate. A full-profile method is only possible when the number of attributes used in the study are limited. In our design this is the case. Furthermore, according to Hair et al. (2010) the full-profile method has several advantages; relatively few judgments per respondent are required, fractional design is possible (each respondent judges a fraction of all possible stimuli), and all attributes are shown simultaneously, which is most realistic. Because of these advantages and the limited amount of attributes used, we chose the full-profile method.
3.4 Design
21
Attribute Level Frequency Actual Ideal Efficiency
OCR 1,5 stars Level deleted
0.4898
0.4985 9541
Third party review 1 out of 10 9 Level deleted Level deleted Level deleted
4 out of 10 9 0.4997 0.9599
1 out of 10 9 0.5006 0.9567
Advertising Level deleted
0.5023
Custom ad 9 0.4991
Price Level deleted
9 0.5064
0.4982
Brand Level deleted
0.4978
0.4871
To ensure that each profile set is shown first just as often as the other profile sets, the order of showing stimuli is randomized for each respondent, to prevent a difference in respondents’ knowledge or concentration for one profile presentation set over the other profile presentation sets. The total number of choice sets shown to each respondent is nine. This acceptable according to Johnson and Orme (1996), who argue that up to twenty tasks can be used without facing a decrease in reliability. In this research was taken a look at keeping the amount of choice sets as low as possible, because it is very important for the research results that the respondents can focus on every choice task set. The more choice sets included, the less focused the respondents are in evaluating the last choices. Nine choice sets is a low amount, so that is good for the reliability of the results.
In choice-based conjoint experiments, often a no-choice (none-) option is included. The main advantage of including a none-option is that a more realistic experiment is obtained (Vermeulen, Goos and Vandebroek, 2008). According to Dhar, (1997) forcing respondents to make a choice between inappropriate choice options might lead to biased parameters and this can lead to wrong conclusions. However, it was decided to not include a none-option in this study design because beer is a well-known product that is bought by consumers frequently. It can be seen as a common purchase. Therefore, respondents will not really feel forced to make a purchase decision, so a none-option was not necessary here.
Furthermore, we choose to use a pictorial presentation of the presentation attributes in this study. A pictorial design presents the stimuli more realistic. Information overload is reduced because respondents don’t have to visualize all the information. Pictorial presentation ensures that homogeneity of perceptions of the presentation attributes is higher among respondents and makes the task itself more interesting for respondents (Green & Srinivasan, 1978). Figure 2 shows how the
22 profile choice sets looked in the survey that was executed. The profile sets were developed by using Microsoft Publisher. Respondents had to make a choice on which crater they would purchase. The entire survey can be found in appendix I.
3.5 Data Collection
To get more insights in consumers online purchase intentions of FMCG, an online questionnaire is conducted, which measures consumer valuations for these five different attributes.
According to Hair et al. (2010) a sample size of minimal 200 is found to provide an acceptable margin of error in the usual applications of CBC Analysis. For identifying segments (latent classes) out of all the respondents, the ideal sample size is at least 200 for each group (Hair et al., 2010). However, although latent classes will be identified in this study, with respect to the feasibility and scope of this study, the current sample size (N=292) is supposed to be acceptable.
In the survey that has been distributed via Qualtrics participated a total of 509 people. Participants were contacted via social media (Facebook, and LinkedIn), personal e-mail, work e-mail and direct begging conversations. Incomplete filled in surveys were excluded manually. This resulted in 292 completed surveys. All respondents received a link, which redirected them to the online survey .The respondents were asked to imagine a situation in which they were doing online groceries and especially about to buy a crater of beer online. Next, before making any decisions or asking any other questions they were redirected to the conjoint analysis and asked to look at the different profiles carefully. The questions were randomized in order to filter out any learning effects. Then, the respondents were asked which of the conjoint profiles they would buy.
3.6 Control Variables
To be able to describe the sample and characterize segments later in this research paper, several control variables are incorporated in this research. After completing the CBC part of the survey, respondents were asked to fill in questions about gender, age, education, profession, income and living area. Besides these demographic questions, questions to measure daily internet usage, monthly online shopping behavior, online review usage and beer consumption behavior were
23 asked after the conjoint part of the survey. These latter questions were asked on a seven-point likert scale ranging from “strongly disagree to strongly agree”. The moderating role of brand importance also has been measured on the same 7-point likert scale. Respondents had to react on the statement that they do not look at a brand when buying beer, in other words, that brand is not important for them / does not play a role when making a purchase decision for beer. This is in line with how Fischer et al. (2010) define brand importance, namely as the role of brands in the customer decision process. With this statement the role of the brand in the customers decision process is measured.
3.7 Segment Solution
In order to achieve the most appropriate number of segments, different decision making factors have to be taken into account. First, we take a look at the information criteria. By looking for the lowest value for the information criterion that was used, the best amount of segments can be determined. There are different sorts of information criteria. Since the absolute fit – in terms of R² and Akaike Information Criteria (AIC) – increases with the number of classes and added variables, it is important to look at the relative fit, which is corrected for the number of model parameters. Therefore, the
Bayesian Information Criteria (BIC) is a better detector of heterogeneity between segments and identifying the optimal
number of classes, since the Log Likelihood is penalized for the number of parameters (Jedidi, Jagpal & DeSarbo, 1997). Therefore segments will be determined by examining the relative fit of the model, measured by the BIC. The lowest BIC score indicates the ideal number of segments.
Besides BIC, classification errors and segment sizes have been taken into account. If there are classes that are very small, this is not very reliable since a few respondents generally cannot represent a population well. We therefore also take this into account. In the end we take a look at the significance of the parameters of the different models and how easy they can be interpreted.
3.8 Covariates Related to Class Membership
There might be additional independent variables (covariates) that have a predictive effect on the dependent variable (choice). These independent variables (covariates) might bring information outside the already available information in the choice tasks to the model to improve the estimation of part-worths (Orme & Johnson, 2009). Active covariates influence the model parameter estimates, while inactive covariates do not. Inactive covariates can be used for describing the classes. Based on the significance levels of these covariates and its effect on overall model fit (BIC) will be determined whether adding active covariates to the model are appropriate.
3.9 Predictive Validity
24
4. RESULTS
In this section we will present the result of the conjoint analysis. The most important demographic characteristics of the respondents of the survey will be presented firstly. After that, the results of the Choice-Based Conjoint (CBC) analysis will be presented, both on aggregated and segment level. Attributes and level preferences are shown. Segments and segment preferences are distinguished. We will conclude with an overview of the hypotheses and evaluate the predictive validity of the model.
4.1 Descriptive Statistics
4.1.1 DemographicsIn this study 292 respondents participated. The questionnaire starts (after a short introduction) with the conjoint choice sets so that respondents will not be biased by prior questions about brand preferences and so forth. After that, some general questions about the respondents were asked that will account for the descriptive below.
Of all the respondents, 169 respondents (58%) are men and 123 respondents (42%) are women. The average age of the respondents is 32 years. The majority of respondents live in rural areas (59%), compared to 41% that live in the city centre. Most of them come from the Eastern part of Holland (52%). About the education of the respondents we can say that 62% of the respondents is highly educated (HBO or WO), only 3% did LBO and 31% did do MBO education. The majority of the respondents have a full time job (48%), but there are also a lot of students (students or working students) in the sample (32%). Around 48% of our respondents earn less than €1500 net, while 22% earns between €1500 and €2500 and 12% earn between €2500 and €3500 net, 9% earns more than €3500 net. Some of the respondents did not want to fill in the question about income, namely 8%. In table 3 below you can find an overview of the descriptive statistics.
Gender Freq. Perc. Education Freq. Perc. Profession Freq. Perc.
Male 169 58 No secundary school 0 0 Student 55 19
Female 123 42 LBO 9 3 Work-student 38 13
MBO 91 31 Unemployed 9 3
HBO 112 38 Part-time empl. 41 14
WO 80 27 Full-time empl. 140 48
Retired 9 3
Total 292 100 Total 292 100 Total 292 100
Location Freq. Perc. Income Freq. Perc. Age Freq Perc.
North of NL 70 24 ≤€ 500 62 21 ≤14 1 0
East of NL 152 52 € 501 ≥ €1.500 82 28 15≥22 50 18
South of NL 9 3 € 1.501 ≥ € 2.500 64 22 23≥30 133 46
West of NL 35 12 € 2.501 ≥ € 3.500 35 12 31≥38 57 20
Mid NL 26 9 € > 3.500 26 9 39≥46 22 8
Rather not say 23 8 47≥54 18 7
≥55 11 4
Total 292 100 Total 292 100 Total 292 100
25 4.1.2 Online Behavior
About internet usage we can conclude from our sample that 34% use the internet less than 4 hours per day, while 26% say they use the internet more than 4 hours per day. The rest of the respondents 40% is in between. Some of our respondents never purchase online (17%), while 57% does less than 2 online purchases per month, 18% does 2-4 online purchases and 8% does more than 4 online purchases per month. Most of the respondents agree or somewhat agree on the statement of making use of online reviews (59%), while 23% does not agree and 18% is neutral here. Also, most of the respondents are open to online groceries. In table 4 below you can see how respondents react on the statements about the trust and irritation reviews give to them. These questions were asked on a 7-point likert scale ranging from strongly disagree (1) to strongly agree (7). A mean of 4 is neutral here. The statement online reviews give me trust scores a mean of 4.58, meaning that the average respondent is more on the side of agreeing on this statement than on disagreeing. On the statement reviews irritate me while shopping online, the mean score is 3.00, meaning that respondents are more on the side of disagreement with this statement.
Reviews give trust Mean SD Reviews Irritate Mean SD
4.58 1.70 3.00 1.69
4.1.3 Beer Consumption Behavior
We also included questions about people’s beer consumption behavior. Most of our respondents prefer Heineken as their favorite beer brand if they have the choice amongst Heineken, Hertog Jan and Grolsch. Namely, 49% of our respondents prefer Heineken, against 28% that have a preference for Hertog Jan and 23% like Grolsch as their favorite brand amongst the three. Grolsch also gets the lowest score from our respondents on the question to rate the beers on a scale from 0 – 100 compared to each other. Grolsch gets an average of 52 out of 100, Hertog Jan gets a 65 out of 100 and Heineken gets the highest average score of 70 out of 100. When our respondents buy a crater of beer they mostly buy it both for themselves and for their guests (52%), 22% buys it solely for themselves and 27% solely for others. From the results we can also see that 19% never buys a crater of beer, 53% buys 1 crater per month, 24% buys a crater per week and 9% buys more than a crater per week.
4.1.4 Brand Importance
Brand importance has been measured on a 7-point likert scale, ranging from strongly disagree (1) to strongly agree (7). On the statement: ''I do not look at brand when I buy a crater of beer'', 60% strongly disagrees. Mean is here 2.04 and SD is 1.70. Meaning most respondents find brand important and therefore on average plays an important role in consumers purchase decisions. 9% agrees on this statement. Meaning for these respondents brand does not play an important role.
26
4.2 Choice-Based Conjoint Analysis - Aggregated Model
For determining what the most appropriate specification for the utility function is, we have to look at how we want to represent our attributes. There are different sorts of preferences functions which are part-worth, quadratic and linear functions. In this conjoint analysis five attributes are included. OCR, third-party reviews, advertisement, price and brand. The attribute brand has nominal levels and can therefore not be estimated as linear. The other four attributes (OCR, third party reviews, advertising and price) are ordinal scaled, so we have to plot them to see if we can treat them as linear (single part worth). We also take information criteria into account in deciding to treat attributes as linear. When estimating the parameters linearly the degrees of freedom increase, because less parameters have to be estimated. The Bayesian Information Criteria (BIC), which is used to determine which model fits best, weights the fit and the parsimony of a model (the lower the BIC, the better the model).If informative criteria goes down by treating attributes as linear this means that the model improves. In figure 3 it can be seen in detail how all the attributes look. We first measured the aggregated model with all the attributes treated as part-worth. The attribute price seems linear and if we measure the model again with price as a single part-worth coefficient (numeric) BIC decreases from 5400.2623 to 5394.6368. We cannot do this for the other attributes, because BIC does not decrease if we try the same for third party recommendations, OCR or advertising.
-0,3 -0,2 -0,1 0 0,1 0,2 0,3
Heineken Hertog Jan Grolsch Brand -0,2 -0,1 0 0,1 0,2 0,3
1,5 stars 3 stars 4,5 stars
27 An overview of the utilities, p-values, ranges and relative importance of the attributes on consumers purchase intentions for FMCG is shown in table 5. All the part-worth of the attribute levels are significantly different from each other, represented by a p-value of 0,000 for all of the attributes.
Attribute Attribute-level Utility p-value Range Rel. imp.
OCR 1.5 stars -0.1328** 0.000 0.3463 16.37%
3 stars -0.0806*
4.5 stars 0.2135**
Third party review 8 out of 10 -0.2272** 0.000 0.4909 23.20%
4 out of 10 -0.0365** 1 out of 10 0.2637* Advertising No ad -0.0770** 0.000 0.1923 9.09% General ad -0.0383 Custom ad 0.1153** Price -0.3145** 0.000 0.629 29.73% Brand Heineken 0.2289** 0.000 0.4575 21.62% Hertog Jan -0.0003 Grolsch -0.2286**
A high online consumer review shows a positive utility (0.2135). A high rated third party review also shows a positive utility (0.2637). High rated OCR, as well as high third party reviews show higher utilities, as compared to the utilities of low OCR (-0.1328) and low third party reviews (-0.2272). Third party reviews shows higher utilities than OCR. Not showing an ad has a negative utility of -0.0770. Showing a general ad has a negative utility of -0.0383. Showing an ad that is customized to the brand shows a positive utility of 0.1153. Price shows the highest negative utility of -0.3145. Brand also shows high positive and high negative utilities. The brand Heineken shows a positive utility of 0.2289. The brand Hertog Jan shows a negative utility of -0.0003. Grolsch has a negative utility of -0.2286.
-0,4 -0,3 -0,2 -0,1 0 0,1 0,2 0,3 0,4 €8,49 €9,24 €9,99 Price
