The Effectiveness of Price Promotions:
It Depends on How the Price Promotion is Framed
Nynke S. van der Broek
Master Thesis MSc Marketing Management Faculty of Economics and Business
University of Groningen July, 2014
UNIVERSITY OF GRONINGEN
FACULTY OF ECONOMICS AND BUSINESS
Master Thesis MSc Marketing Management
The Effectiveness of Price Promotions:
It Depends on How the Price Promotion is Framed
Author: Nynke S. van der Broek s1765418
Camperstaat 16 II 1091 AE Amsterdam nynkevdbroek@gmail.com +31(0) 6 - 22 87 37 76 1st supervisor: dr. J.E.M. van Nierop 2nd supervisor: dr. ir. M.J. Gijsenberg
July, 2014
ABSTRACT
Previous research about price promotion framings is limited by estimating the effect of the framings on consumers’ perceptions. This thesis is an extension on previous research and studies the short-term sales effect of different price promotion framings in the personal care segment. Based on A.C. Nielsen data of the largest Dutch drug retailer, the effect of different price promotion framings on short-term sales volume will be compared. It can be concluded that different price promotion framings differ in terms of effectiveness. The multi-item price promotion framing which is framed with a free product on top of the purchases, displayed as “X + Y” is the most effective price promotion framing, in relation to a percentage off, cents off or multi-item promotion framed as “X for €Y”. This study provides useful marketing implications for the price promotion strategy of manufacturers in the personal care segment.
Keywords: price promotion, price promotion framing, promotion effectiveness, A.C. Nielsen data
ACKNOWLEDGEMENT
In 2008 I started my Bachelor Business. I soon became interested in the field of Marketing. After completing my Bachelor Business and having a great board year at the Marketing Association Rijksuniversiteit Groningen (MARUG), I decided to apply for the Master Marketing. To this day, marketing has always found my interest. The college years have yielded a big part of my marketing knowledge.
Now, that I am writing the last sentence of my thesis, I realize that I have to bring this knowledge into practice and explore the world of marketing myself.
I would like to thank dr. E.J. van Nierop for providing me with useful feedback and valuable suggestions at every stage of this MSc Marketing Thesis. I would like to thank dr. ir. M.J. Gijsenberg for being willing to serve as a second supervisor and motivate me to improve this thesis. I would also like to thank my fellow thesis group members for providing me with helpful feedback.
Besides that I would like to thank xx and my colleagues for providing me the data and making it possible to combine my thesis with my new fulltime job.
A special thanks goes out to my family and friends for their support during my study and this thesis process. I would like to express my greatest gratitude to my parents Jan and Anke van der Broek and Sander Schuitema who have supported me unconditionally during my study and in every choice I made.
Amsterdam, July 2014
Nynke van der Broek
TABLE OF CONTENTS
1. INTRODUCTION 6
1.1PROBLEM STATEMENT 6
1.1.1MANAGERIAL RELEVANCE 6
1.1.2LITERATURE RELEVANCE 7
1.2STRUCTURE OF THE THESIS 9
2. THEORETICAL FRAMEWORK 10
2.1BACKGROUND 10
2.2DEVELOPING HYPOTHESES 11
2.2.1PRICE PROMOTIONS IN GENERAL 11
2.2.2PRICE PROMOTION FRAMINGS 14
2.2.3PRICE PROMOTION FRAMINGS AT DIFFERENT DISCOUNT LEVELS 16
3. METHODOLOGY 17
3.1DATA 17
3.2VARIABLES DESCRIPTIONS 19
3.2.1DEPENDENT VARIABLE INCREMENTAL SHORT-TERM SALES VOLUME 19
3.2.2PREDICTOR VARIABLES 20
3.2.3CONTROL VARIABLES 20
3.3MODEL 22
3.4STATISTICAL ANALYSIS PLAN 24
4. RESULTS 25
4.1DESCRIPTIVE STATISTICS 25
4.2HYPOTHESES AND RESEARCH QUESTION TESTING 27
4.2.130%DISCOUNT VERSUS 50%DISCOUNT 28
4.2.2MULTI-ITEM VERSUS SINGLE-ITEM PROMOTION 29
4.2.3FOUR DIFFERENT FRAMINGS 29
4.2.4FRAMINGS PER DISCOUNT LEVEL 31
4.3CONTROL VARIABLES 32
5. DISCUSSION 34
5.1THEORETICAL IMPLICATIONS 36
5.2MANAGERIAL IMPLICATIONS 37
6. LIMITATIONS & FURTHER RESEARCH 39
7. CONCLUSION 42
REFERENCES 43
APPENDIX A EXAMPLE OF A CROSS CATEGORY PROMOTION
AND A NON-CROSS CATEGORY PROMOTION 48
APPENDIX B GRAPHICAL REPRESENTATION OF USED DATA 49
APPENDIX C CORRELATION 50!
1. INTRODUCTION
‘Two deodorants for the price of one’, ‘Now, your favourite shower gel with 50%
off’. Price promotions seem to be the norm in the Dutch personal care segment.
Companies spend large amounts of money to influence consumers’ support of their brands with price promotions. Promotion pressure and promotion frequency increased enormously in the last years. A.C. Nielsen data showed that the promotion pressure had increased with 8% from 52% to 60%1 in the Dutch bulk personal care market from 2011 until the beginning of 2014.
1.1 Problem Statement
Manufacturers and retailers are faced with different decision-making problems when placing their products in promotion. One of these problems is how to communicate a price promotion in order to increase the sales volume. This study will examine different price promotion framings and their short-term sales effect.
1.1.1 Managerial Relevance
Due to the rise of private label products, competition in the personal care market increases. According to Reijnders and Luijten (2014) consumers experience less distinction between private label and national brands, as they are increasingly opting for the cheapest brand. This results in a decrease in revenue for these national brands. National brands of the personal care segment are constantly in promotion, resulting in a continuing subsidization of loyal customers of these brands. It however also promotes switching behaviour amongst consumers which in turn results in a declining brand and store loyalty.
A large part of the marketing budget of national brands is being used for price promotions, which causes other marketing activities, such as innovations, to be postponed. Considering the importance of such other marketing activities, savings need to be made in the price promotion budgets. In order to avoid consumers switching to competitors however, the price promotions strategies need to increase their effectiveness (Reijnders & Luijten, 2014). As the increasing
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
1!Based on A.C. Nielsen data of the Dutch personal care segment in the Drug bulk sector. The promotion pressure of 2011 is the mean of week 4 till week 52 of 2011. The promotion pressure of the beginning of 2014 is the mean of week1 till week 16 of 2014.
promotion pressure is not sustainable for national brands, considerable attention has to be given to the effectiveness of price promotions. This thesis will contribute to the development of an effective price promotion strategy.
1.1.2 Literature Relevance
Previous research focused mostly on promotion frequency (Kalwani & Yim, 1992; Kumar & Pereira, 1995), promotion depth (Kalwani & Yim, 1992), scheduling of price promotions (Kumar & Pereira, 1995), competitive reactions of price promotions (Leeflang & Wittink, 1992) and the effect of including the reference price in a promotion frame (Blair & London, 1981; Chandrashekaran &
Japgpal, 1995). The frequency of price promotions increases, affecting customers buying behavior to such an extent that they ‘learn’ to buy their products only during a promotion. The promotion pressure has a positive effect on the short- term sales volume. However, this pressure is at such a high level that the promotion budget of managers will be overspent (Krishna, Currim &
Schoemaker, 1991). Therefore it is even more important to create an effective promotion strategy which includes all the aspects of a promotion.
The lack of research attention to price promotion framings is surprising.
Early literature about price promotion framing is limited to consumers’ perception of different price promotion framings. Della Bitta, Monroe and McGinnis (1981) demonstrated the effect of different price promotion framing on consumers’
perceived offer value and concluded that price promotions which are presented in dollar terms are more positively valued compared to price promotions framed in percentage off. In contrast, Hardesty and Bearden (2003), DelVecchio, Krishnan and Smith (2007) as well as Chen, Monroe and Lou (1998) found that low price products with percentage off discounts are perceived as deals which provide a larger saving than deals that are framed as cents off. Wansink, Kent and Hoch (1998) concluded that multi-item promotions are more effective than percentage and cents off promotions. In addition, Das (1992) also suggested that different framings have different impacts. He concluded that framings like ‘2 for $X’, and
‘buy 1 get 1 at $X’, are most effective. In contrast, Krishna, Briesch, Lehmann and Yuan (2002) conducted a meta-analysis of five different price framings and concluded that there was no difference in effect between promotions framed in
dollars, percentage, multi-item promotions, coupon promotions or promotions with a premium.
This thesis is a complement to previous research regarding price promotions and will focus on different promotion framings. Previous research about price promotion framings shows the effect of the price expectations, choice and reference price (see e.g. Della Bitta, et al., 1981; Delvecchio et al., 2007;
Hardesty & Bearden, 2003; Chen et al., 1998; Laroche, Pons, Zgolli, Cervellon &
Kim, 2003; Das, 1992; Sinha & Smith, 2000; Krishna et al., 2002). This study focuses on the effect of price promotion framings on short-term sales response and differs from previous research by measuring the effect of the price promotion framings by using weekly A.C. Nielsen scanner data. The benefit of A.C. Nielsen data is that the data allows for investigation of actual consumer behavior. By identifying the effects of the promotion framing, marketing managers may get more insight in how to optimize their price promotion strategies. The following main research question results from this problem:
Which price promotion framing is most effective to engage an effective promotion strategy?
To allocate the best promotion attributes, companies can create an optimal promotion strategy to increase promotion pressure of their products. Based on A.C. Nielsen data the effectiveness of various promotions will be examined. This study focuses on the short-term sales responses of different price promotion framings in the personal care segment and encompasses the Dutch market. Four different product categories2: deodorant, hair care, shower gel and hair styling (bulk categories) of a national brand were analysed in a period between 2011 till the beginning of 2014. These products are frequently purchased products, also known as fast moving consumer goods (FMCG). Different price promotion framings are examined: the framing “X + Y” and “X for €Y” which belong to the multi-item promotion category and the percentage off and cents of framing which belong to the single-item promotion category. Multi-item promotion is a promotion framing where the consumer receives one or more extra product(s) on
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
2 The definition of category is used as defined by marketing managers and A.C. Nielsen.
top of their purchase(s). Single-item promotions give a discount on one single unit.
The purpose of this study is to provide recommendations on how manufacturers may set up an effective sales promotion strategy in terms of price promotion framing to realize more savings in order to “beat” the promotion pressure. In spending these savings on other marketing activities and product innovations, companies may strengthen their market position. Note: as this thesis focuses on consumer price promotions, trade promotions are not considered.
1.2 Structure of the Thesis
The remainder of this thesis will proceed accordingly: the following section briefly reviews the relevant literature of price promotions in general. It will also give more in-depth attention to price promotion framings, which will lead to the empirical questions and hypotheses. The third chapter elaborates on the methodology of the analysis and contains a description of the used data, presentation of the model and the statistical analysis plan. The results of this study will be given in the fourth section. In section five the results will be discussed and the answer of the problem statement can be found. Finally, limitations and directions for future research will be provided as well as a brief conclusion.
2. THEORETICAL FRAMEWORK
In this chapter, the main concept of this study will be discussed. In Figure 1 the conceptual model of this thesis is given. This conceptual model reflects the main issues that will be discussed. First, price promotion in general will be considered and then a more in-depth review of price promotion framings will be given by using previous research. Based on the findings of previous research the hypotheses and empirical questions will be formulated.
FIGURE 1
Conceptual Framework
2.1 Background
Research shows that price promotions result in a substantial initial sales increase (for reviews see e.g., Blattberg, Briesch & Fox, 1995; Nijs, Dekimpe, Steenkamp
& Hanssens, 2001; Pauwels, Hanssens & Siddarth, 2002; Gedenk, Neslin &
Ailawadi, 2006; Blattberg & Neslin, 1990; Van Heerde, Leeflang & Wittink, 2004; Mela, Gupta & Lehmann, 1997).3 Generally speaking, price promotions account for almost two-thirds of the promotion budget, but unfortunately only 16% is profitable (for a review see Blattberg & Neslin, 1990). Marketing managers understandably are concerned with the allocation of scarce marketing resources for the purpose of improving the sales performance of their products
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
3 The effect of price promotion in the long-run will be considered as negligible in this thesis, to avoid misleadingly post-promotion effects (see e.g. Blattberg & Jeuland, 1981), as a category in the personal care segment finds itself, on average, once in every four weeks in promotion.
Short-term Sales Volume Price Promotion Frame
Empirical question 1
Hypotheses 1, 2 and 3
Empirical question 2 Multi-item
X + Y X for €Y
Single-item Percentage off Cents off
Price Promotion Depth
(Dekime, Hanssens, Silva-Risso, 1999). As a result, industry experts are calling for effective and cost efficient promotion strategies.
2.2 Developing Hypotheses
The purpose of this thesis is to investigate which price promotion framing of a national brand is most effective. In this chapter a literature overview of the promotions is provided. First, a general overview of price promotions is presented. Secondly, a more detailed discussion regarding price promotion framings will be given and the related empirical questions and hypotheses will be formulated.
2.2.1 Price Promotions in General
Because price promotions can be easily implemented, they are considered as an interesting marketing instrument by marketing managers (Leeflang, Wittink, Nedel & Naert, 2000; Blattberg & Neslin, 1990). Price promotions are temporary price discounts offered to consumers and are the most common form of sales promotions (e.g. Blatteberg & Neslin, 1990; Blattberg & Neslin, 1989; Kumar &
Pereira, 1995; Darke & Chung, 2005; Gedenk et al., 2006: 345). Price promotions are used because they have an immediate effect on the short-term sales and they increase the familiarity with the brand while stimulating consumption (Dekimpe, Hanssens & Silva-Risso, 1999; Mela et al., 1997; Pauwels et al., 2002; Raju, Srinivasan & Lal, 1990). Short-term sales effects can be measured by short-term sales volume of the products. Short-term sales volume can be defined as the sales volume that is generated in the promotion week offered by marketers or intermediates in the promoting store or other selling point. The incremental sales volume is related to a price promotion. If the price promotion had occurred in that week, this incremental volume would not have been generated. (Abraham &
Lodish, 1993) An increase in the short-term sales volume can result from brand switching, category switching, purchase of extra quantity (penetration) and purchase acceleration which in turn may be prompted by promotions. Purchase acceleration refers to the behaviour of the consumer when they buy a product earlier than usual (Gedenk et al., 2006: 345-348; Blattberg & Neslin, 1990: 353) and causes higher inventory levels (e.g. Gedenk et al., 2006). Due to the promotional price cut, consumers will buy more units of the promoted product
than in the absence of such a promotion or they will adjust their timing of such a purchase (Blattberg & Neslin, 1989).
Consumers make multi-category decisions in their shopping trips. A price promotion in one category effects the consumers’ purchase intensity in related product categories. (Machanda, Ansari & Gupta, 1999; Mulhern & Leone, 1991) However, the cross-promotion effects are smaller than in the promoted category.
Machanda et al. (1999) found that this cross category effect does not occur for complementary product sales. Almost no literature was found about the impact of cross category promotions; considering that cross category promotions have more impact on sales than promotions which are related to one category. However, Hruschka, Lukanowicz & Buchta (1999) concluded if promoting one category has a positive effect on another category, it would obviously be an advantage to promote both categories simultaneously.
The counter effect of price promotions is that consumers’ reference prices decrease (Chandrashekaran & Jagpal, 1995; Gedenk et al., 2006: 349). When a brand is promoted often the consumer becomes reluctant to pay the non- promotional price for a product. In addition, this also lowers the perceived quality of the product by the consumer (Kalwani & Yim, 1992; Kahn & Louie, 1990;
Pauwels et al., 2002). Similarly, Mela et al. (1997) found that sales promotions increase price sensitivity and destroy brand equity in the long run. Another counter effect is that price promotion can lead to deceleration: consumers will now wait until a product is in promotion. Mela, Jedidi and Bowman (1998) found in their study that when consumers expect a promotion in the foreseeable future, they will buy less products in the period leading up to that (expected) promotion.
However, when the same product is on sale, consumers tend to buy more products. This situation is consistent with the deceleration effect.
Price promotions can be supported by several marketing instruments (see Blattberg & Nelsin, 1989). A distinction in marketing instruments can be made between supportive (marketing instruments with focus on the price cut) and non- price promotions (marketing activities with focus on the product). Supportive promotions4 are communication instruments enhancing consumers’ awareness of
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
4 Supportive marketing activities have a direct result on sales (e.g. Kumar & Pereira, 1995).
Because this study focuses on the relation between price promotion and short-term sales, only supportive marketing instruments will be discussed in this section.
price promotion, like features, displays, premiums and promotion packaging.
Non-price promotions like sampling and events, on the other hand, focus on a brand or product instead of focusing on the price cut.
Price promotions with feature can be described as an announcement of the price promotion in a retailer’s flyer, which as a result, will attract consumers to the store. A display is in-store activity and attracts customers to the product when they are in the store. (Kumar & Pereira, 1995) Displays influence the point of sales (POS) of consumers and gives extra attention to products. They are individually placed shelves which increase the store shelf space of products. This will positively influence the sales. (Narasimham, Neslin & Sen, 1996)
When elasticity of marketing activities (change in sales of the promoted product specified in percentage, induced by a given percentage change in the markting budget) on sales is positive; an increase in supportive marketing activities leads to higher sales (Narasimham et al., 1996; Van Heerde, Leeflang &
Wittink, 2002; Gedenk et al., 2006: 348). Narasimham et al. (1996) found that an unsupported price discount of 15% generates an increase in sales of 34%, whereas by adding a feature or display to the price promotion the sales response will increase with respectively 161% and 293%.
A promotion can have different stages of promotion depths. Promotions depth can be described as the average size of discount to which consumers are exposed (Nijs et al., 2001; Della Bitta et al, 1981). Della Bitta et al. (1981) compared different promotion framings with the consumers’ perception of saving.
According to them there are no significant differences in the consumers’
perception of savings on price promotions with 30%, 40% or 50% discount.
However, there is significant difference between a 10% and 30-50% discount level. In addition, Van Heerde, Leeflang and Wittink (2001) concluded that price promotions have a threshold effect. Price promotions lower than 10% do not lead to an increase in short-term sales. Besides, they found a saturation effect at a level of 25%. Price promotions higher than 25% do not lead to additional short-term sales. The results of Grewal, Marmorstein and Sharma (1996) indicate that consumers’ processing of information increases when the discount of a promotion rises from a low level to a moderate level. When the price reduction increases further within a plausible level, the information processing of the consumer declines. They made a distinction between price promotions with low, moderate
and high cents off. Respectively these cents off promotions had a price cut of 17%, 29% and 50% of the reference price. In this study, the observed national brand offers only promotions with around 30% and 50% discount, in the observed period. In response to previous research the following empirical question (1) is formulated:
Is the effect of a 30% discount equal to the effect of a 50% discount on short-term sales volume?
2.2.2 Price Promotion Framings
In their study, Hardesty and Bearden (2003) mentioned that both retailers and manufacturers should be interested in the conditions under which consumers value the different types of price promotion framings. Different price promotion framings can be employed to communicate the same economic quantity discount, for example 1 + 1 and 50% off. Framing the same information in different ways can have a great impact on consumer decision-making (Chen et al., 1998; Della Bitta et al., 1981; Gedenk et al., 2006: 350).
There are different types of price promotions which can be divided into two different price promotion framings: multi-item price promotions and single- item price promotions. The differences between multi-item and single-item promotions lies within the quantity of products that have to be purchased to receive the discount.
Multi-item promotion is a promotion framing where the consumer receives one or more extra product(s) on top of their purchases (e.g. buy two, get one free;
two for one; buy three for €Y). This framing is used to encourage the consumer to buy the quantity that is suggested by the multiple units (e.g. Gedenk et al., 2006:
347; Blattberg & Neslin, 1990: 350). In other words: to receive the discount, the requirement is to buy more than one product. In practice most of these promotions are related to two or more categories to stimulate cross category selling.
Single-item promotions, which only give a discount on one single unit, refer to percentage off and cents off promotions. To receive the discount, it is not required to buy more than one product. Percentage off promotions are promotions where the price reduction is presented in percentage terms (e.g. 50% off) (e.g.
Chen et al., 1998; Hardesty & Bearden, 2003; Gedenk et al., 2006: 350). Cents off is a reduction in monetary terms (e.g. with €2 off!).
Manning and Sprott (2007) investigated multi-item promotions and its effects on the quantity of purchase intention and found a positive effect of this type of promotion. Foubert and Gijsbrechts (2007) concluded that the amount of products which are needed to receive the promotion, decreased after a certain level. The relation between the framed quantity of a multi-item framing and the consumers’ purchase quantity shows an inverted U. Multi-item promotions with low quantities were ineffective in increasing consumers’ purchase quantity for all observed super market products and they found no difference in effectiveness between an eight-item promotion and a twenty-item promotion. In previous research evidence is found that multi-item promotions have a higher effect on short-term sales compared to single-item promotions. This finding was however, limited to the price promotion framing “X for €Y” (Manning & Sprott, 2007;
Foubert & Gijsbrechts, 2007; Wansink et al., 1998). Wansink et al. (1998) found that the sales of stores which used a multi-item promotion, were on average, 32%, higher than the stores which used a single-item promotion in nine of the thirteen observed categories. Based on previous research the following hypothesis (1) is formulated:
H1: The effect of multi-item promotions on short-term sales volume is higher than the effect of single-item promotions on short-term sales volume.
As mentioned before, Wansink et al. (1998) performed limited research of one specific multi-item promotion “X for €Y”. They gave no attention to multi-item price promotions framed as “X + Y”. Yet these different multi-item price promotions may have different effects. Multi-item promotions framed like “2 + 1”
do not mention a price cut, but emphasise an additional product as a free gift.
Getting products as free gifts can affect the perceived value of the promotion. The promotion is framed as gains in the perception of the consumer. (Diamond &
Campbell, 1989; Kim & Kramer, 2006) Multi-item promotions displayed like “3 for €5”, emphasise no free products in their framing. In the observed data only the multi-item framings “X + Y” and “X for €Y” were found. Therefore, the following hypothesis (2) is formulated:
H2: The effect of multi-item promotions framed with a free product “X + Y” on short-term sales volume is higher than the effect of multi-item promotions framed
without a free product “X for €Y” on short-term sales volume.
Della Bitta et al. (1981) demonstrated the effect of different price promotion framings on consumers’ perceived offer value and stated that price promotions, which are presented in dollar terms, are more positively valued compared to price promotions framed in percentage off. In contrast, Hardesty and Bearden (2003) DelVecchio et al. (2007) and Chen et al. (1998) found that low price products with percentage off discounts are perceived as providing a larger saving than deals that are framed as cents off. Therefore, the following hypothesis (3) is formulated:
H3: The effect of single-item promotions with a percentage off framing on short- term sales volume is higher than the effect of single-item promotions with cents off
framing on short-term sales volume.
2.2.3 Price Promotion Framings at Different Discount Levels
In the section about price promotion in general, it was questioned if the effect of a promotion with a discount of 30% will be equal to a discount of 50%. This empirical question was based on the findings of Van Heerde et al. (2001), Grewal et al. (1996) and Della Bitta et al. (1981). They found a saturation effect around a discount level of 30%: no differences were found between promotions around the 30% discount level and a 50% discount level. In this study, the focus is on price promotion framings; therefore it is even more interesting to look at the different framings and to question whether this assumption is true for all the observed framings or not. In response to this, the following empirical question (2) is formulated:
If the effect of a 30% discount is equal to a 50% discount level, will this be the case for all the price promotion framings or does it depend on how the price
promotion is framed?
3. METHODOLOGY
This chapter will first describe the used data. Subsequently, the variables that are adopted in the model will be discussed, whereupon the model will be presented.
Finally the statistical analysis plan will be given.
3.1 Data
A.C. Nielsen5 scanner data was used for this analysis. Data of four different product categories in the bulk personal care segment were analysed. The categories are: deodorant, shower gel, hair care and hair styling. The categories consist of multiple stock-keeping units (SKU’s), varying from purpose (e.g.
prevent dry skin, for sensitive skin, prevent heavy sweating, protect coloured hair), smell or volume (amount of millilitres). These product categories are frequently purchased products and have similar price levels, varying from €2.29 to €5.69 (see Table 1, p. 19). The prices are the same across all the stores of the retailer. The prices of the products within the categories hardly fluctuated in the observed period. The few price changes that occurred, fell within the price range, displayed in Table 1 (p. 19).
Analysis was performed with A.C. Nielsen scanner data (aggregated on retail level) from a time span of 169 weeks: starting in week 4 of 2011 up to and including week 16 of 2014. Only the weeks with a promotion were adopted in the analysis, which made a total of 468 weeks. The focus is on category sales when analysing the effect of the different price promotion framings.
Data is drawn from the largest drug retailer in the Netherlands which has approximately 840 stores. The retailer sells national brand and private label products. The retailer had weekly feature promotions in the personal care segment; therefore, weekly data was used. A price promotion lasted either one or two weeks. Each category was in promotion at least twice a month.
The A.C. Nielsen scanner data showed in which week a feature was presented by pointing out what the weighted distribution was in a specific week of
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
5 A.C. Nielsen is a company that obtains purchase data of retailers. The data provide information on (among other things) purchases by category, brand, premiums and sampling. In this study, A.C.
Nielsen data of the baseline sales, incremental sales, price index, weighted distribution of the feature and the promotion share of the sales volume were used.
that feature. A.C. Nielsen defines the weighted distribution of a flyer as follows:
the weighted distribution indicates how many stores were supported with a flyer in the observed week, displayed in percentage of the total stores (whereby 100%
indicate that all stores where supported with a flyer). A store is considered as
“supported by a flyer” when the flyer is distributed in the geographic target consumer area.6 In each observed promotion weeks the promotion was featured in the flyer, whereby the weighted distribution was higher than 85%.
Flyers of week 4 of 2011 until week 16 of 2014 were used to obtain information about which price promotion framing was used in the promotion week and to gain information about the length of the promotion period. The price promotions featured in the flyer were offered at all the stores of the retailer. All price promotion framings used, can be described as comparative price promotions and are displayed in a feature. Non-featured promotions cannot be found in the data. Therefore, the effect of a feature cannot be measured in this study.
Price promotions on one SKU are not discussed in this thesis, because in these promotion weeks the total products that were sold in promotion were less than 20% of the total category sales. Promotions related to a whole category had a promotion share of 64% or higher which indicates that the 64% (or more) of the purchased products were bought in promotion in the observed week.7 To be able to compare different promotion weeks, only the category promotions were adopted in the study and the promotions on one SKU were removed, because these promotions have too little impact on the incremental short-term sales of a whole category.
In this study the effect of single-item promotions (percentage off and cents off) and multi-item promotions (“X + Y” and “X for €Y”) were analysed. The cents off framings were displayed as “with €X discount” or as “from €X, now
€Y”. These two cents off framings were taken together, because there was not enough data to measure the effect of the two different framings. An overview of the amount of weeks that are found with a price promotion framing is shown per category in Table 1. In Table 1 (row 2) the minimum quantity requirements per price promotion framing can also be found. This is the number of products that
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
6,7Nielsen, Handleiding Scantrack, prepared by Nielsen, January 2013.
have to be bought to receive the discount. These quantities are used in all the four categories.
TABLE 1
Number of Promotion Framings per Category
Price
range X + Y X for €Y Percen-
tage off Cents off Tot. wks in promotion Quantity
requirements 2, 3 or 4 2 or 3 1 1
Deodorant €2.29 -
€5.69 93 15 3 20 131
Shower gel €2.99 -
€5.45 86 11 6 11 114
Hair care €3.45 -
€2.55 101 13 1 2 117
Hair styling €2.69 -
€4.59 92 9 1 4 106
3.2 Variables Descriptions
The variable incremental sales volume as percentage of the baseline sales measured per category was adopted in the model as dependent variable. The prediction variables are the price promotion framings and price index. The variables: category, summer, cross category and second week will be adopted as control variables. The meaning of the variables will be described below. Note, all the variables are dummy variables.
3.2.1 Dependent Variable Incremental Short-term Sales Volume
Incremental short-term sales volume per category as percentage of the baselines as a result of price promotion was measured for four categories of one brand at one retailer chain. Baseline sales are the sales that would have occurred for the brand in a respective category in the given week if the promotion had not taken place. Baseline sales are affected by distribution, regular price of the products and non-supportive price promotion advertisement. Price promotions, displays and features affect incremental sales volume (for a review see e.g. Blattberg & Neslin, 1990; Kumar & Pereira, 1995). By analysing the incremental sales volume in percentage of the baseline sales of the different categories, it was possible to
analyse the four categories together. Besides that, by measuring the effects on sales volume, it makes it possible to compare sales across different price and discount levels.
3.2.2 Predictor Variables
Price Promotion Framing. Price promotions, which are adopted in this study, were single-item promotion framed as cents off (1 euro discount and 2 euro discount) or percentage off (30% and 50%) and multi-item discounts framed as
“X + Y” (“1 + 1”, “2 + 1”, “2 + 2” and “3 + 1”) or “X for €Y” (2 for €5, 3 for €5, 2 for €6 and 3 for €6). Most of the multi-item promotions are cross category promotions. Only price promotions related to one or more entire categories are included in this model. Price promotions on only one SKU within a category are not discussed in this thesis, because these promotions have a promotion share of less than 20% of the total sales, where other promotions have a promotion share of 64% or higher8. All price promotions are displayed in a feature ad.
Price Index. The price of the products in a category is not equal and varies over time. To compare the depth of discount levels between the categories, a price index is entered in the model. The price index is adopted as one minus the discount rate; a promotion of 30% gave a price index of 1 - .3 = .7. Price index has a value of .7 (30% discount) or .5 (50% discount). These discount rates do not entirely correspond with the actual discounts given. In order to compare the effect of the price promotion framings, the discount is rounded off. For example the 2 + 1 framing gives the consumer a discount of 33%. The price index will be: 1- .3 = .7.
3.2.3 Control Variables9
Summer10. According to Blattberg and Nelsin (1990: 244-250) variation in sales volume may occur due to seasonally influences. Data was obtained over more than three years, which means that all seasons (summer, autumn, winter and spring) occurred in the data set. It may be expected that
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
8 A.C. Nielsen defines the promotion share as follows: the share of products that was purchased in promotion, displayed in percentage from the total product sales (Handleiding Scantrack, prepared by Nielsen, January 2013).
9 Analyses with the variables display and premium were performed, but these variables had no significant effect, therefore they were removed from the model.!
10!Analyses with the variables autumn, winter and spring were also performed, but these variables had no significant effect, therefore this study only focus on the season summer.!
summer generates more sales compared to the other seasons. People sweat more and therefore odds are that the need for more personal care products will increase during the summer, therefore the variable summer is adopted in this model.
Second Week of the Promotion. In the observed period, there were one-week and two-week promotions. Previous research found that the duration of a promotion has an effect on sales (Van Heerde et al., 2000, 2004). It might be expected that the first week of the promotion generates more sales response compared to the second week of the promotion.
Cross Category Effect. A cross category price promotion refers not only to one relevant product category but to two or more product categories. In the observed period, there were both promotions found which only related to one category as well as promotions which related to more than one category. For example a promotion framed as “1 + 1” applied to both deodorant and shower gel.
This means that the consumer can buy one deodorant and one shower gel to receive the discount. It was expected that cross category promotions generate more sales response per category compared to promotions which focus on merely one category. This can be explained by the fact that cross category promotions stimulate cross selling. This expectation was based on prior literature, which found that a promotion in one category will lead to more sales in related categories (Mulhern & Leone, 1991, for a short overview of more previous research see section 2.2.1). Therefore, the expectation is that promotions in more categories will also stimulate cross selling. However, in this thesis no attention is paid to sales volume that is generated in a related category. Only attention is paid to the category which is observed in the considered promotion week. To include the variable cross category, it can be measured if consumers buy more products in cross category promotions or not. In other words: does a cross category promotion generate more sales in the considered category? To be clear what a cross category promotion is, in Appendix A, a cross category promotion (Figure 1) and a non cross category promotion (a promotion that only refers to one category) (Figure 2) can be found.
Category. Data of four product categories are used. The categories are deodorant, shower gel, hair care and hair styling. The products of these categories are bulk products. They are fast moving consumer goods (FMCG), which means that the products are nondurable goods and frequently purchased by consumers.
!
In previous studies about price promotions, lead and lag effect were taken into account (e.g. Van Heerde et al., 2000; Van Heerde et al., 2004). Lead effects result from purchase deceleration (pre-promotional dip) and lag effects arise through consumers’ stockpiling (acceleration) (Blattberg & Nelsin, 1990;
Leeflang et al., 2000). Consumers are not buying the same product after the promotion ends, because of the additional products that they bought during the promotion (Blattberg & Neslin, 1990). This possible effect is not included in the model. In the observed period, it occurred several times that after a promotion week a new promotion started for the same product category. It was not possible to estimate the lead and lag effects, because the pre- and post promotion dips will interfere with promotion weeks. Therefore, the model in this thesis will not take lead and lag effects into account. Only attention was given to promotion weeks.
Other weeks were not adopted in the model. Therefore no time series analysis is done.
All the promotions were announced in a feature ad, therefore the variable feature was not adopted in the model. Other variables were not considered in the model, because including all theoretically relevant variables can result in too many parameters with insufficient observations and multicollinearity. This will lead to higher standard errors for the estimated coefficients.
3.3 Model
The focus is on incremental short-term sales volume per category, when analysing the effectiveness of the price promotion framings to adopt retailers’ perspective.
This variable is adopted as dependent variable in the model. The different framings and the variable price index were adopted as predictor variables. The model shows all possible combinations of the interaction effects of framing on price index. There will be controlled for the variables summer, cross category, category and second week. Only price promotion weeks were analysed. Hence, the model can be specified as follows:
! Si=β0+ β1 j
j=1 3
∑ Framingi+β2PriceIndexi+ ( β3k
k=1 3
∑ Framingi⋅ PriceIndexi)
+β4Summeri+β5SecWki+β6CrossCati+ β7l l=1
3
∑ Categoryi
!
for i is a particular given promotion week
Si incremental short-term sales volume in % of the baseline sales.
Framingji dummies for the price promotion framings in a promotion week i, j {0,…,3}, with j = 0: Cents off, j = 1: X + Y, j = 2: X for €Y and j = 3: Percentage off.
PriceIndexli which is a dummy for the price index in a promotion week i, taking a value of 1 if the price index is .5, and a 0 if the price index is .7.
The relation between price index and short-term sales volume in percentage of the base line sales is moderated by framing, k {0,.., 3}, with k = 0: Price index * Cents off , k = 1: Price index * X + Y, k = 2: Price index * X for Y and k = 3: Price index * Percentage off.
Summeri which is a dummy for the summer in a promotion week i, taking a value of 1 if the week falls in the summer, and a 0 if the week falls in the season autumn, winter or spring.
SecWk i which is a dummy for second week in a promotion week i, taking a value of 1 if the promotion falls in the second week, and a 0 if the promotion falls in the first week.
CrossCati which is a dummy for a promotion refers to more than one category in a promotion week i, taking a value of 1 if the promotion is a cross category promotion and taking a value of 0 if the promotion is not a cross category promotion.
Categoryki dummies of the different categories in a promotion week i, l {0,…,3}, with l = 0: Hair styling, l = 1: Deodorant, l = 2: Shower gel and l = 3: Hair care.
∈
∈
∈
3.4 Statistical Analysis Plan
In order to test the hypotheses and to provide answers to the empirical questions, a multiple linear regression was conducted in SPSS11. The model described in section 3.3 will be adapted to test each individual hypotheses and research question separately. First a linear regression was done to measure the direct effect of the price index on the incremental short-term sales volume to answer the first empirical question. Second, the effect of multi-item and single-item promotions will be measured by using a linear regression. A third linear regression was done to test the direct effect of the different framings. To answer the second empirical question, the moderating effect of framings on price index was measured by also using a linear regression. In all the analyses, there will be controlled for the variables summer, second week, cross category and category.
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
11! SPSS for Mac, Rel. 22.0.0. 2013. Chicago: SPSS Inc.!
4. RESULTS
This chapter describes the results on the study of the effect of price promotion framings and the relation between price index (a given discount) and the incremental short-term sales volume as percentage of the base line sales. First, the descriptive statistics of the used data will be given. Second, the effects of a 30%
and 50% promotion will be measured. Third, multi-item versus single-item price promotion will be estimated. Fourth, the different framings: “X + Y”, “X for €Y”, cents off and percentage off will be estimated and after that the effect of these framings as moderating effect on the relation between price index and incremental sales volume will be measured. These analyses will be done by using a multiple linear regression whereby there will be controlled for the variables summer, second week, cross category and category.
4.1 Descriptive Statistics
Figure 2, 3, 4 and 5 present a graphical representation of the incremental sales in volume and the given discount of the observed period of the categories:
deodorant, shower gel, hair care and hair styling are given. The height of the bars indicate the discount percentage and the line demonstrates the incremental sales per week. The graphs show that sales increase when discount is given. Interesting to note however, is that it does not show that the higher the discount, the higher the incremental sales. Therefore the suggestion can be made that the effect of discount is moderated by one or more variables, this can indicate that the framing of the discount can play a role. The figures also show the non-promotion weeks.
These weeks were removed from the data when analysing the hypotheses and empirical questions.
In Table 2 an overview can be found of the price promotion framings which were used by the retailer in the observed period. As can be seen in the table, the mean of the framing “X + Y” at a discount of 50% is the highest. This involves a framing of “1 + 1” or “2 + 2”. In addition, also the framing “X + Y” generates a higher value compared to other framings with a 30% discount. As complement to this table, in Appendix B, a scatterplot can be found which represents the
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60 80 100 120 140 160 180 200
1 52 103 154
Discount Increm. Sales x1000
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60 80 100 120 140
1 52 103 154
Discount
Increm. Sales x1000
FIGURE 2
Incremental Sales Volume and Discount given per Week of Category Deodorant
!
FIGURE 3
Incremental Sales Volume and Discount given per Week of Category Shower gel
FIGURE 4
Incremental Sales Volume and Discount given per Week of Category Hair care
FIGURE 5
Incremental Sales Volume and Discount given per Week of Category Hair styling
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 10 20 30 40 50 60 70
1 52 103 154
Discount
Increm. Sales x1000 0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 20 40 60 80 100 120 140 160
1 52 103 154
Discount
Increm. Sales x1000
incremental sales per discount level. The plot shows that a discount level of 50%
has the highest incremental peaks compared to a discount level of 30%.
TABLE 2
Means of Incremental Sales in % of the Baseline Sales per Promotion
Incremt. sales
volume in % (x100)
!
Mean (SD) 30% discount X +Y 1.99 (1.29)
!
X for €Y 1.40 (1.28)
! Percentage off .42 (.41)
!! Cents off 1.37 (.94)
50% discount X +Y 4.34 (2.83)
!
X for €Y .87 (1.51)
!
Percentage off 1.08 (.76)
Cents off 1.22 (1.21)
Note. N = 468 !
Where Table 2 shows the means of the different price promotion framings per discount level, Table 1 in Appendix C represents the means, standard deviations and the zero-order Pearson correlation of all the variables that are included in the model. Because all independent variables are dummy variables, the means of these variables lies between the zero and one.
4.2 Hypotheses and Research Question Testing
In chapter 3 the model can be found which is used to test the hypotheses and to provide answers to the empirical questions. This model is adapted to test the specific empirical questions and hypotheses12. In the next sections, for each specific empirical question and hypothesis, the description of the adjusted model can be found. In each model there will be controlled for the variables summer, second week, cross category and category. These variables have a significant
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
12 When testing the linear regression models the assumption that the error term has to be homoscedastic is not met for the overall model and the adjusted models. Applying a log transformation to the data did not sufficiently overcome this problem. Interpreting the outcomes might be obscured due to this limitation in the data.
!
