Thesis title: Cross-Channel Synergies In A Multichannel Strategy
Author: Chiel van Til
Faculty: Economics and Business Department: Marketing
Completion date: June 19th, 2016
Supervisor: Peter van Eck Second Supervisor: Alec Minnema
Organization: University of Groningen
Address: Kremersheerd 158 9737 PE Groningen E-mail: c.van.til@student.rug.nl Student number: S2808145
MASTER THESIS OPTIMIZING MEDIA STRATEGIES
‘’C
ROSS
-‐
CHANNEL SYNERGIES IN A MULTICHANNEL STRATEGY
’’
Abstract
The goal of this paper is to get a better understanding of the synergetic effects on sales of the exposure to YouTube, in combination with Television and pre-‐roll advertising via the Dutch website of RTL. A dataset containing a sample of households and their exposure to the three marketing tools mentioned above was used to test the hypothesis.
In contrast to my expectations, both YouTube and TV had the significant positive effect on sales. The same applied for the impact of online-‐ and offline marketing tools which both had a significant positive effect on sales. In line with my expectations, TV advertisement had a higher effect on sales than pre-‐roll advertising via the website of RTL. Statistically significant positive synergetic effects on sales were found between both YouTube & TV and YouTube & RTL.
The results of this paper suggest that synergy effects forBrand X were only derived from the impact on sales, thus Brand X should focus on this. Another point for Brand X would be to focus more attention on YouTube as advertising source, as it has shown to be effective both individual and together with other marketing tools. I found that synergies are complex and it is not certain that they will be achieved. Literature therefore suggests using an Integrated Marketing Communication program, which would make all marketing tools focus on the same message and the same objective.
Preface
Table of content
1. Introduction ... 4
2. Literature Review ... 5
2.1 Effectiveness of stand-alone advertising media ... 5
2.2 YouTube as an advertising channel: ... 5
2.3 Using more than one medium to advertise: ... 6
2.4 Synergy effects: ... 7
3. Conceptual Model ... 9
4. Research method ... 10
4.1 Exploring the dataset ... 10
4.2 Model specification ... 13
5. Results ... 14
6. Conclusion ... 23
Managerial implications ... 28
Limitations and further research ... 29
Literature ... 30
1. Introduction
In 2014, the world has spent $1.6 trillion on media and this number is likely to rise to $2.1 trillion in 2019 (CMO Council, 2016). As said in the early 1900’s by John Wanamaker ‘’half the money I spend on advertising is wasted; the trouble is I don’t know which half’’(Wanamaker, around 1900), whereby he already stressed the importance to make marketing more accountable for. This is also underlined by academics such as Verhoef and Leeflang (2009) who argue that marketing needs to regain influence in the boardroom. Furthermore, it is an important topic for the Marketing Science Institute (2016).
To make marketing more accountable for, marketing productivity has to be measured, which can be done by several steps according to Kumar et al (2004), who states that one of the steps is making tactical actions, such as improving advertising to increase the marketing assets. Improving individual channels can be used to improve advertising effectiveness, but also using several channels to communicate the same message (e.g. print advertising and radio) might improve marketing productivity even more (Naik and Raman, 2003). This is called synergy, a goodness of fit between multiple channels (Birgelen et al, 2006).
Research on the synergies between online- and offline marketing channels has been conducted in several articles, whereby studies examined the synergies between TV, Radio, Magazines and for example Banners (Ha, 2008), (Schibrowsky et al, 2007), (Assael, 2011). However, to my knowledge, there has been no study that addresses the synergy question between several ‘’traditional’’ marketing methods and new marketing channels like content communities, as for example YouTube in a multichannel campaign. This paper addresses this gap in research and will focus on the synergy between the ‘’traditional’’ marketing methods and advertising on YouTube.
Therefore, the research question in this paper is: ’How does Advertising on YouTube increase chance of purchase in a multichannel strategy?’
of the first 13 weeks of 2014 whereby the purchase behavior and exposure to marketing tools of Brand X is measured. The paper is structured as follows: first, an outline of the theoretical foundations will be given before deriving to the hypothesis, second, the methodological foundations and findings will be described and finally, the implications for practice and future research will be summarized.
2. Literature Review
First, the known effectiveness of stand-alone advertising media will be given whereby YouTube will be discussed separately as it is the medium, which will have the main focus in this paper. Second, reasons for using more than one medium to advertise are given. Third, a definition a synergy effect, known synergy effects in research and possible explanations for synergy effects are given.
2.1 Effectiveness of stand-alone advertising media
Danaher and Dagger (2013) argue that Social Media advertising does not influence the offline purchase probability and only leads to higher traffics to the retailer’s websites, thus not significantly influencing the purchase probability. Other stand-alone dvertising channels such as direct mail, television, radio and newspapers did significantly influence sales (Danaher and Dagger, 2013). Campbell et al (2011) agrees and argues that consumer ads on YouTube might be considered to be ‘’noise’’ and thus not higher the purchase probability. Next to the effects of Social Media advertising as a stand-alone medium, other effects of offline- and online channels have been researched before. Joel (2009) found that television is still one of the most popular media to advertise on and is not using its effectiveness over time. According to earlier studies of (Bruner & Kumar, 2000), (Dreze & Hussherr, 2003), (Pavlou &Stewart, 2000), (Briggs et al, 2005), advertising effects are generally similar -or even stronger- online and offline as consumers’ reactions on Internet advertising are often similar to offline advertising (Spilker-Attig and Brettel, 2010). Abrahams (2009) also supports this by showing that in store sales increase when advertising online. But newer studies have shown that digital media as stand-alone advertising source are less effective than offline media (Danaher, 2011).
2.2 YouTube as an advertising channel:
YouTube’s revenue was 4 billion dollars. This revenue is mainly gathered by advertising. YouTube’s advertising consists of commercials before the video's, banners, final call to action, call to action overlay and vloggers (Yoganarasimhan, 2011), (Siu, 2014). Companies choose to advertise on YouTube as advertising via YouTube increases brand recognition (Li and Lo, 2015) and also as it enhances peoples brand attitudes when used in a multichannel campaign (Loewenstein et al, 2011). Other reasons are that online advertising is still considered to be low cost (Naik and Peters, 2009) and that online advertising boosts sales in general and can outweigh the costs of advertising by seven times (Lewis and Reiley, 2014), (Rutz and Bucklin, 2010). YouTube is also the best Social Medium to drive conversion according to AOL (2015), who analyzed 500 million clicks, 15 million conversions and three billion impressions.
Looking at the literature about the stand-alone effectiveness of advertising channels, the following hypothesis was derived:
H1: Offline stand-alone channels have a bigger positive effect on offline purchase than online stand-alone advertising.
H1a: TV campaigns have a higher effect on offline purchases than YouTube. H1b: TV campaigns have a higher effect on offline purchases than Website advertisement.
H1c: Offline campaigns have a higher effect on offline purchases than online advertisement.
2.3 Using more than one medium to advertise:
use several media for the same campaign as accounting for media synergies significantly improves the forecasting power for product sales (Lin et al, 2013).
2.4 Synergy effects:
As shown in the previous part, one reason to use several media for the same campaign would be due to synergy effects. Synergy effects can be defined as the total effect of media channels combined is larger than the sum of its individual parts (Naik & Peters, 2009), also explained as 1+1=3 (Romaniuk, 2013) and by the old Greek philosophe Aristotle ‘’the whole is bigger than the sum of its parts’’. Synergy effects are not only found within offline- or online channels, but also between these channels (Dinner et al, 2014).
There are several explanations found in literature that try to explain synergy effects. In their article, Stammerjohann et al (2005) explain synergy effects via three ways. The first explanation would be the encoding variability theory, which states that when a message is received via multiple media, the message will have a stronger and clearer information network in the brain (Tavassoli, 1998). The second explanation would be the repetition-variation theory, which is explained by Gibson (1996), which states that more exposures to an advertisement are better than only one exposure. More exposures, in combination with variety of promotional tools and harmony could according to Gibson (1996) create a positive attitude toward a brand. The third explanation would be the principles of selective attention. Kahneman (1973) states that individuals pay most attention to stimuli that are complex and familiar or which are simple and novel whereby repeating a message across different channels would lead people to become more familiar with the message. The more attention someone pays to a certain message, the more effective this message becomes.
When looking to the explanations given in the article by Stammerjohann et al (2005), it becomes clear that a synergy effect consists out of several channels spreading the same advertising message as it will be more effective than using one single channel. As in the case of my example, YouTube spreads an advertising message that is shared by other channels (e.g. Television and Website advertisements), a synergy effect might be found here as well.
media types and banner advertising for higher attention and credibility, but did not manage to find synergies that have an impact on purchase intentions. Also other researchers could not find synergies; Havenla et al (2007) and Dijkstra et al (2005) did not find synergy effects between traditional and online advertising when it comes to purchase intentions, just increased cognitive responds compared with internet only campaigns.
Authors Media Synergy
Found
Sales | Purchase Probability as DV?
Abraham (2008) On- and offline Yes Sales
Jagpal (1981) Radio and Newspaper Yes Sales
Naik and Raman (2003)
TV and print Yes Sales
Danaher and Dagger (2013)
Catalogues, TV, DM, Radio, Internet, Social Media
Yes Sales
Godfrey et al (2011) Telephone, E-mail and mail Yes, negative
Purchase Probability
Chang and Thorson (2007)
TV and Web Yes Purchase Probability
Table 1: Research about Synergy Effects with Sales or Purchase Probability as DV.
As shown in the table above, synergy effects are researched and found in literature. Researchers found synergies between the different media channels such as TV, Radio, Newspapers, Social Media and Banner Advertising. All the synergy effects above were found with Sales or Purchase Probability as the dependent variable.
2015). According to Yizao and Rigoberto (2016), successful usage of Social Media will also increase awareness of a brand, which would lead to greater probability of purchase. Other online advertising are also known to create synergies as showed by Abraham (2009) who found synergies between several online channels such as banner advertising and online search ads. Thus, according to the known literature, advertising on Social Media and advertising online might lead to synergies in combination with other (online) media. Therefore, I expect that synergies between the channels used in our data (Television and Website advertisement) will also be established in combination with spreading the same advertising message via YouTube. Therefore, the following hypothesis is derived:
H2: There is a positive synergy effect of advertising between:
H2a: YouTube and website advertisement on offline purchases H2b: YouTube and TV campaigns on offline purchases
H2c: YouTube, advertisements on websites and TV campaigns on offline purchases
3. Conceptual Model
Image 1: Conceptual Model
4. Research method
This chapter explains the research method. First, the dataset will be described and secondly, the models used will be presented.
4.1 Exploring the dataset
The database consists of panel data of Brand X of 10703 households for week 1 till 13 of 2014. The database was collected and provided by the research company GFK and contains data on household’s purchases, demographics of the household’s and exposure to Brand X’s advertisements.
Purchase data:
The data about the purchase history gives us insights in what day people bought Brand X, for what price, how many units, about the price per unit, whether the unit was for sale and whether people bought products at a competitor as well.
Demographics:
was measured in 7 categories and income, which was measured categorical from less than 700 euro to 4100 euro or more in 17 categories. Each of these variables will be added to the tests as control variables; they might influence the variables that I will measure.
Exposures:
The dataset also contained the number of exposures to three different media channels. There are two online variables and one offline variable, of which the online variables are based on actual contacts:
- TV Exposures: Measures the amount of times that people were exposed to a TV advertisement of Brand X.
- RTL Website: Measures the amount of times that people were exposed to a Pre-roll advertisement of Brand X showed on the website of a Dutch cable TV company called RTL.
- YouTube: Measures the amount of times that people were exposed to a Pre-roll advertisement of Brand X showed at YouTube.com.
Purchase data
In the observational period of 13 weeks, offline purchases were made. purchases were Brand X products and were made at competitors, which can also be seen in graph 1. The total sales of Brand X were calculated by price per unit * the number of units bought, which gives a total sales of €. As one can see, the sales are pretty stable over time, which is usual for a FMCG like Brand X.
Graph 1 Number of Purchases
Advertisement
A table was created to show in what week Brand X used each media channel for advertisement. As seen in table 2, the first five weeks, Brand X only used TV campaigns to send out their promotional message. From week six onward, this message is also spread via YouTube advertising and in week six, seven, ten and eleven, Brand X also choose to use pre-roll advertisement on the website of RTL. There was no week without any advertisement. With the analysis below, we will see whether the extra advertising exposure led to extra sales.
Table 2 Advertisement Per Week
On average, people were exposed to the advertising of Brand X via the television .4798 times per week, whereby the minimum is zero and the maximum amount of exposures is eight. This amount is way lower for both RTL and YouTube, where people are exposed to .0136 times per week (YouTube) and .0038 times per week (RTL) on average.
N Minimum Maximum Mean Std. Deviation
YouTube 12425 ,00 2,00 ,0136 ,12389
RTL 12425 ,00 1,00 ,0038 ,06139
TV 12425 ,00 8,00 ,4798 ,92617
Buyers and non-buyers
To get a quick view whether the exposure to advertising of Brand X made people purchase more in the first place, an independent sample t-test was performed. With this independent sample t-test, the mean numbers of exposures to advertising of Brand X of purchasers of Brand X were compared with the non-purchasers. Results of this test showed that the two groups did not statistically significantly differ in exposure to the advertising of Brand X via YouTube, RTL and TV. Only in one case, people that purchases were marginally significantly more exposed to YouTube advertisement than people that did not purchase. This means that the Households in this dataset that purchased a product were not exposed more often to advertising than people who did not purchased a product.
Equality of Variance t Df Sig. (2-tailed) Mean difference Largest mean YouTube Not assumed 1.801 1414.77 .072 .00813 Equal RTL Assumed -.342 12423 .732 .732 Equal TV Assumed .699 12423 .484 .484 Equal
Table 4 Differences in exposures between buyers of Brand X and non-buyers.
Synergetic effects
The dataset of GFK did not contain any variables that could measure the impact of the synergetic effects on sales or purchase probability. Therefore, I created these variables in order to test my hypothesis.
4.2 Model specification
I used two dependent variables to measure the effect of advertisement on sales of Brand X. These variables are a binomial variable that covers whether a Household ID bought (1=Y) or did not bought (0=N) a product of Brand X in a certain week and the sum of the sales. The binominal variable was used to measure hypothesis two, where the sum of sales was used for hypothesis one and two. The sum of sales per week per Household ID ranged between € 0.00 to € 77.39.
I estimated nine models:
H1c: Sales = α + β1Agei +β2EDUi +β3Ini + β4YTit + β5RTLit + β6TVit
H2a: Sales = α + β1Agei +β2EDUi +β3Ini + β4YTit + β5RTLit + β6TVit + β7IntYou*RTLit
H2a: Pr (Yi=1) = (exp(β1Agei +β2EDUi +β3Ini + β4YTit + β5RTLit + β6TVit + β7IntYou*RTLit))/ (1+ exp (β1Agei +β2EDUi +β3Ini + β4YTit + β5RTLit + β6TVit + β7IntYou*RTLit)
H2b: Lag_Sales = α + β1Agei +β2EDUi +β3Ini + β4YTit + β5RTLit + β6TVit + β7IntYou*TVit H2b: Logit_Pr (Yi=1) = (exp(β1Agei + β2EDUi + β3Ini + β4YTit + β5RTLit + β6TVit +
β7IntYou*TVit))/ (1+ exp (β1Agei +β2EDUi +β3Ini + β4YTit + β5RTLit + β6TVit + β7IntYou*TVit) H2c: Sales = α + β1Agei +β2EDUi +β3Ini + β4YTit + β5RTLit + β6TVit + β7IntYou*RTL*TVit H2c: Pr (Yi=1) = (exp(β1Agei +β2EDUi +β3Ini + β4YTit + β5RTLit + β6TVit +
β7IntYou*RTL*TVit))/ (1+ exp (β1Agei +β2EDUi +β3Ini + β4YTit + β5RTLit + β6TVit +
β7IntYou*RTL*TVit)
Where:
t= The week number ranging from Week 1 – Week 13. i= The HouseHold ID ranging from ID 1723 – ID 800133.
𝑈sales = The sales of a product of Brand X for household i, in Week number t;
Pit= Probability of buying a product of Brand X for household i, in week t
α = Constant | Intercept;
β = Parameter
TVit= The amount of exposures to Television Advertisement for household i, in week t.
YTit= The amount of exposures to YouTube Advertisement for household i, in week t.
RTLit= The amount of exposures to RTL Advertisement for household i, in week t.
LSit= Variable for Lagged Sales for household i in week t.
Ini= Income for Household ID i.
Agei = The age of the Housewife of Household ID i.
EDUi= The level of education of the Housewife of Household ID i.
Intit*Intit= Interaction between the advertising sources where int= YouTube*RTL, YouTube*TV,RTL*TV or YouTube*RTL*TV, i is the household and t is the week.
5. Results
Multicollinearity
synergies showed any VIF scores higher than 1.040 which is lower than the cut-off value of 4 which means moderate multicollinearity and definitely lower than 10, which means strong multicollinearity (Malhotra, 2010). Therefore, I could continue with the analysis without removing any variables.
H1a: ‘’TV campaigns on itself have a higher effect on offline purchases than YouTube’’
For the first part of this hypothesis, I want to find out how much variance in sales is explained by the exposure of household’s to TV advertisement and YouTube advertisement. Sales are measured in scale, and therefore, a multiple linear regression was calculated to predict Sales of Brand X on Age, Education, Income, YouTube contacts and TV contacts. A significant regression equation was found (F(5,7788)= 9.782, p<.000), with an adjusted R2 of .006, meaning that the model explains 0.6% of all sales which is quiet low. YouTube contacts (B=1.119) and TV contacts (B=.111) were found significant. All results can be found in table 4. Thus, for every extra exposure to YouTube, the sales of Brand X per Household go up with € 1,119. The sales of Brand X also increase with € 0,111 per additional TV exposure. In order to test the hypothesis that TV advertisement (b=.036) and YouTube’s advertisement (b=.049) standardized beta weights are statistically significantly different from each other, their corresponding 95% confidence intervals were estimated via bias corrected bootstrap (1.000 re-samples). In the event that the confidence intervals overlapped by less than 50%, the beta weights would be considered statistically significantly different from each other (p<.05) (Cumming, 2009). As can be seen in Table 6, there appeared to be an overlap in the confidence intervals. To evaluate the hypothesis more precisely, the average overlap was calculated from the confidence intervals of both YouTube and TV (.027), which is 59% of the total confidence interval of YouTube. This indicates an overlap of >50% between the two variables, which according to Cumming’s method (2009) should not be considered statistically significant. Thus, we cannot accept H1, as there was no significant difference. We can only conclude that the standardized Beta of YouTube advertisement is numerically higher than the standardized Beta of TV.
Education ,054 ,034 ,018 1,602 ,109
Income ,036 ,008 ,048 4,256 ,000
YouTube Contacts 1,119 ,260 ,049 4,311 ,000
TV Contacts ,111 ,035 ,036 3,164 ,002
Table 5 Adj. R2=.006 p=<.000
Lower Limit Point Upper Limit
TV .032 .047 .095
YouTube .013 .035 .059
Table 6 Upper and Lower limits of the Confidence Intervals
H1b: ‘’TV campaigns on itself have a higher effect on offline purchases than Website
advertisement’’
For the second part of this hypothesis, we want to find out how much variance in sales is explained by the exposure of household’s to TV advertisement and RTL Website advertisement. Therefore, a multiple linear regression was calculated to predict Sales of Brand X on Age, Education, Income, RTL contacts and TV contacts. A significant regression equation was found (F(5,7788)= 6.358, p<.000), with an adjusted R2 of .003, which means that the model explains 0.3% of all sales. All results can be found in table 7. TV contacts (B=.108) was found significant. RTL contacts were found to be insignificant and thus we can conclude that exposure to TV advertisements is a better predictor of Sales than exposures to RTL pre-roll advertisements. This means that within this model, one more exposure to TV could lead to €0,108 more sales. Therefore, we can accept H1b, TV is a better stand-alone advertising channel than the RTL website.
Predictors
Unstandardized Coefficients Standardized Coefficients
t p-Value
B Std. Error Beta
(Constant) ,062 ,255 ,244 ,808
Age -,013 ,016 -,009 -,802 ,422
Income ,037 ,008 ,050 4,392 ,000
TV Contacts ,108 ,035 ,035 3,086 ,002
RTL Contacts ,653 ,528 ,014 1,236 ,216
Table 7 Adj. R2=.003 p=<.000
H1c: ‘’Offline campaigns itself have a higher effect on offline purchases than online
advertisement’’
As one can see above, sometimes offline campaigns are better than online campaigns, but it can also be the other way around. Therefore, I tested H1c as well. A multiple linear regression was calculated to predict Sales of Brand X on Age, Education, Income, Online Advertising (sum of RTL and YouTube exposures) and Offline Advertising (TV exposures). A significant regression equation was found (F(5,7788)= 9.899, p<.000), with an adjusted R2 of .006, thus explaining 0.6% of the sales. All results can be found in table 8. Online advertising (B=1.011) and Offline advertising (B=.111) were found significant. Within this regression, one more exposure to offline advertisement increases sales with €0,111 and one more exposure to online advertisement increases sales with €1.011. In order to test the hypothesis that Offline advertisement (b=.036) and Online advertisements (b=.050) standardized beta weights are statistically significantly different from each other, their corresponding 95% confidence intervals were estimated via bias corrected bootstrap (1.000 re-samples). In the event that the confidence intervals overlapped by less than 50%, the beta weights would be considered statistically significantly different from each other (p<.05) (Cumming, 2009). As can be seen in Table 9, there appeared to be an almost full overlap in the confidence intervals. To evaluate the hypothesis more precisely, the average overlap was calculated from the confidence intervals of both Online- and Offline advertisement (.03475), which is 77% of the total confidence interval of the Offline advertisement. This indicates an overlap of >50% between the two variables, which according to Cumming’s method (2009) should not be considered statistically significant. Thus, we can not accept hypothesis h1c, as they are not significantly different from each other. The only conclusion that we can make is that the standardized Beta of Online advertising is numerically higher than the standardized Beta of Offline advertising.
(Constant) -,010 ,255 -,038 ,970 Age -,010 ,016 -,007 -,642 ,521 Education ,054 ,034 ,018 1,611 ,107 Income ,036 ,008 ,049 4,287 ,000 Offline Advertising ,111 ,035 ,036 3,180 ,001 Online Advertising 1,011 ,231 ,050 4,378 ,000 Table 8 Adj. R2=.006 p=<.000
Lower Point Upper
Online .007 .047 .101
Offline .013 .035 .058
Table 9 Upper and Lower Limits of the Confidence Intervals
H2a_Linear: ‘’There is a positive synergy effect of advertising between YouTube and website
advertisement on offline purchases ‘’
Graph 1 Effect of the Standardized Beta's
Predictors Unstandardized Coefficients Standardized Coefficients t p-Value B Std. Error Beta (Constant) ,044 ,253 ,174 ,861 Age -,012 ,015 -,009 -,760 ,447 Education ,048 ,033 ,016 1,437 ,151 Income ,036 ,008 ,048 4,301 ,000 TV Contacts ,112 ,035 ,036 3,227 ,001 RTL Contacts -,449 ,532 -,010 -,845 ,398 YouTube Contact ,868 ,258 ,038 3,360 ,001 YouTube*RTL Contacts 33,992 3,010 ,129 11,292 ,000 Table 10 Adj. R2=.022 p=<.000
H2a_Logistic ‘’There is a positive synergy effect of advertising between YouTube and
website advertisement on offline purchases ‘’
A logistic regression analysis was conducted to predict the purchase probability for Brand X for 7794 observations of households using Age, Education, Income, YouTube contacts, RTL contacts, TV contacts and an interaction between YouTube and RTL as predictors. A test of the full model against a constant-only model was statistically significant, indicating that the predictors as a set reliably distinguished between people who bought Brand X and people who
did not bought Brand X (X2 (7) = 62.088, p=.000). Nagelskerke’s R of .017 indicated a weak relationship between the prediction and grouping.
Prediction success overall was 55.5% (55.2% for no purchase and 57.6% for purchase at a cut-off point of .099), which means that it correctly classified 55.2% of the non-purchases and 57.6% of the purchases. The Wald criterion demonstrated that the number of YouTube contacts and TV contacts made a significant contribution to the prediction of sales for Brand X as shown in table 11. Exp(B) values indicated that when the number of exposures to YouTube advertisements is raised by one unit, the odds ratio is 1.896 times as large and therefore people are 1.896 times more likely to purchase a product of Brand X. Exp(B) values also indicated that when the number of TV exposures is raised by one unit, the odds ratio is 1.125 times higher and thus people are 1.125 times more likely to purchase Brand X. Unfortunately, the interaction effect was not significant (p=.205) and thus H2a was rejected based on the impact on purchase probability.
Predictors: B S.E. Wald Sig. Exp(B)
Age -,046 ,018 7,006 ,008 ,955 Education ,004 ,037 ,010 ,919 1,004 Income ,055 ,009 35,205 ,000 1,057 YouTube Contacts ,640 ,219 8,499 ,004 1,896 RTL Contacts -,430 ,734 ,344 ,558 ,650 TV Contacts ,117 ,036 10,498 ,001 1,125
RTL Contacts * YouTube Contacts 2,036 1,608 1,604 ,205 7,663
Constant -2,442 ,282 75,144 ,000 ,087
Table 11 Nagelkerke's R=.017 p=.000
H2b_Linear: ‘’ There is a positive synergy effect of advertising between YouTube and TV
For the second part of this hypothesis, we want to find out how much variance in sales is explained by the exposure of household’s to interaction between TV advertisement and YouTube advertisement. Therefore, a multiple linear regression was calculated to predict Lag_Sales of Brand X based on Education, Age, Income, Number of TV contacts, Number of YouTube contacts, Number of RTL contacts and interaction effect between YouTube and TV. All results can be found in table 13. A significant regression equation was found (F(7, 7187) = 6.638, p<.000, with an adjusted R2 of .005, explaining 0,5% of the variance in sales. In this regression, YouTube Contacts (B=.787), TV Contacts (B=.075) and Interaction between YouTube and RTL (B=.10.442) were found significant. Thus, within this model, an additional exposure to YouTube leads to €0,787 extra sales, an additional exposure to TV leads to €0,075 and an exposure to both YouTube and TV leads to an additional €10,442 extra sales. As the interaction between YouTube and TV is significant and has a positive standardized beta of .041, we can accept H2b ‘’There is a positive synergy effect of advertising on offline purchases between YouTube and TV campaigns’’. So in this case, 1+1 again equals three, as there is a positive synergy effect as shown in graph 4 and 5. Thus, we can accept H2b based on the impact on sales.
Graph 4 Effect of the Standardized Beta’s
Predictor Unstandardized Coefficients Standardized Coefficients t p-Value B Std. Error Beta (Constant) ,048 ,267 ,181 ,856 Age -,008 ,016 -,006 -,464 ,643 Education ,048 ,035 ,016 1,360 ,174 Income ,037 ,009 ,049 4,110 ,000 TV Contacts ,075 ,038 ,023 1,972 ,049 RTL Contacts -,370 ,531 -,008 -,696 ,486 YouTube Contacts ,787 ,262 ,036 2,999 ,003 YouTube*TV Contacts 10,442 3,007 ,041 3,473 ,001 Table 12 Adj. R2=.005 p=<.000
H2b_Logistic ‘’ There is a positive synergy effect of advertising between YouTube and TV
campaigns on offline purchases’’
A logistic regression analysis was conducted to predict the purchase probability for Brand X for 7794 observations of households using Age, Education, Income, YouTube contacts, RTL contacts, TV contacts and an interaction between YouTube and RV as predictors. A test of the full model against a constant-only model was statistically significant, indicating that the predictors as a set reliably distinguished between people who bought Brand X and people who did not bought Brand X (X2(7) = 60.717, p=.000). Nagelskerke’s R of .016 indicated a weak relationship between the prediction and grouping. The results can be found in table 14.
number of TV exposures is raised by one unit, the odds ratio is 1.123 times higher and thus people are 1.123 times more likely to purchase Brand X. Unfortunately, the interaction effect was not significant (p=.707) and thus H2b was rejected based on the purchase probability.
Predictors B S.E. Wald p-Value Exp(B)
Age -,046 ,018 6,997 ,008 ,955 Education ,003 ,037 ,009 ,927 1,003 Income ,055 ,009 35,181 ,000 1,057 YouTube Contacts ,640 ,235 7,429 ,006 1,896 RTL Contacts -,106 ,610 ,030 ,863 ,900 TV Contacts ,116 ,036 10,122 ,001 1,123
TV Contacts * YouTube Contacts ,122 ,325 ,141 ,707 1,130
Constant -2,440 ,282 75,001 ,000 ,087
Table 13 Nagelkerke R2=.016 p=.000
H2c: ‘’ There is a positive synergy effect of advertising between YouTube, advertisements on
websites and TV campaigns on offline purchases’’
As shown above, several synergies have been established in this dataset between YouTube advertisement, RTL advertisement and TV advertisement. Next step is to test hypothesis H2c, where we look for a synergetic effect between the three mentioned channels. None of the households were exposed to all three media types in one day and thus a weekly aggregated version of the dataset was created. But still, even in the weekly aggregated version, no household ID was exposed to all three types of media in the same week. Using lag variables to increase the number of households exposed to all three types of media did not give any new findings. Thus, unfortunately, this hypothesis could not be tested.
6. Conclusion
chance of purchase in a multichannel strategy?’. Several hypotheses were specified to find an answer to this question and they were tested using a GFK dataset on household level and three types of advertisement namely YouTube, RTL website and TV advertisement. All answers to the hypothesis can be found below.
First, hypothesis 1a ‘’ ‘’TV campaigns on itself have a higher effect on offline purchases than YouTube’’ was answered. In this study, both variables had a positive effect on purchases but were not statistically significantly different from each other, which makes them equally important. The only difference being that YouTube had a numerically higher effect on purchases than TV. Thus, this hypothesis could not be accepted. Second, hypothesis 1b ‘’TV campaigns on itself have a higher effect on offline purchases than Website advertisement’’ was answered. TV advertisement was the only statistically significant predictor and therefore it had a bigger effect on offline purchases than Website advertisement. Hence, this hypothesis was accepted. Third, hypothesis 1c ‘’Offline campaigns itself have a higher effect on offline purchases than online advertisement’’ was answered. This hypothesis had to be rejected as, although both variables had a significantly positively effect on offline purchases, they were not statistically significantly different from each other. Thus, both online and offline advertisement had the same amount of impact on offline purchases and YouTube and TV advertisement have the highest individual impact on sales.
Fourth, hypothesis 2a ‘’There is a positive synergy effect of advertising between YouTube and website advertisement on offline purchases ‘’ was answered via both offline sales and purchase probability. A statistically significant positive synergy effect was achieved between YouTube and Website advertisement based on sales. There was no synergy effect found based on the purchase probability, as the interaction between YouTube and Website advertisement was statistically insignificant. Therefore, based on the impact on sales, hypothesis 2a was accepted. On the other hand based on the impact on purchase probability, hypothesis 2a was rejected. Fifth, hypothesis 2b ‘’There is a positive synergy effect of advertising between
YouTube and TV campaigns on offline purchases’’ was answered. The synergy effect existed based on the impact on sales and was statistically significant. Again, based on purchase probability, the interaction between the YouTube and TV campaigns was not statistically significant. Hence, based on the impact on sales, hypothesis 2b was accepted, but based on the impact on purchase probability it was rejected. Finally, hypothesis 2c ‘’ There is a positive
on offline purchases’’ was answered. I assumed a positive synergy effect of advertising between the three advertising channels YouTube, website advertisement and TV advertisement on offline purchases. This hypothesis was not measured due to the insufficient number of households that were exposed to all three advertising channels (n=0).
Thus, to answer the main question, advertisement with YouTube can increase the chance of purchase both individually and in combination with other advertising channels. Individually, YouTube also has a positive impact on both sales and on purchase probability. In combination with other channels, YouTube achieves synergy effects in combination with other advertising channels and together, they have a positive impact on sales.
Discussion
In this paragraph, the results of the established hypothesis will be discussed per hypothesis.
Hypothesis: Supported? Result:
H1a: ‘’TV campaigns on itself have a higher effect
on offline purchases than YouTube’’
Rejected. YouTube seems to have a higher
numerical effect on offline purchases than TV, but it is not significant. H1b: ‘’TV campaigns on itself have a higher effect
on offline purchases than Website advertisement’’
Accepted. TV advertisement is the only
significant predictor. H1c: ‘’Offline campaigns itself have a higher effect
on offline purchases than online advertisement’’
Rejected. Online Advertisement has a higher numerical effect, but it is not significant.
H2a: ‘’There is a positive synergy effect of
advertising between YouTube and website advertisement on offline purchases ‘’
Accepted. A significant positive synergy effect.
H2a: ‘’There is a positive synergy effect of
advertising between YouTube and website advertisement on offline purchases ‘’
Rejected. The interaction was not significant.
H2b: ‘’ There is a positive synergy effect of
advertising between YouTube and TV campaigns on offline purchases’’
Accepted. A significant positive synergy effect.
H2b: ‘’ There is a positive synergy effect of
advertising between YouTube and TV campaigns on offline purchases’’
Rejected The interaction was not significant.
H2c: ‘’ There is a positive synergy effect of
advertising between YouTube, advertisements on websites and TV campaigns on offline purchases’’
Could not be tested
Hypothesis 1a ‘’TV campaigns on itself have a higher effect on offline purchases than YouTube’’ was rejected as TV campaigns and YouTube advertisement have the same impact on sales. Several reasons might have led to this result. First, YouTube commercials are only counted as being ‘’viewed’’ if the whole commercial is shown (YouTube, 2015). Thus, meaning that only people who are at least ‘’interested’’ in Brand X are been counted as exposed, which makes YouTube more effective. Second, YouTube’s performance might be influenced by the ease at which a Brand X can target people via the Internet. Unlike TV, which works like a shotgun, YouTube is able to precisely target certain contextual, behavioral or demographical features, working like a sniper (Adweek, 2015). Finally, it is known that Brand X is a brand that is mainly consumed by youngsters (Sloot, 2016), which is confirmed by the negative age linked to the purchase probability in the results of table 11 and table 13. Youngsters are more likely to use YouTube and less likely to watch TV and Rice et al (1998) already stated that marketing communications are more effective when the advertisement message is communicated to consumers via a channel that is preferred by them.
Hypothesis 1b ‘’TV campaigns on itself have a higher effect on offline purchases than Website advertisement’’ was accepted as TV advertisement was the only significant predictor. Pre-roll advertisement on the RTL website was ineffective during this campaign which can be explained by several reasons. First, Brand X products in this dataset are all purchased offline, which according to Danaher and Dagger (2013) is not good for online marketing channels. Second, the amount of exposure did not took into account the quality of the advertisements which could be high in the case of YouTube and low in the case of the pre-rolls on the RTL website (Winer, 2009). Third, the targeting could be ineffective in the case of RTL. Finally, when measuring only a single channel might not fully account for the total impact of the advertisement as shown by Dinner et al (2014).
and Spilker-Attig (2010), who showed that online media would get more powerful and would be equally important as offline advertisement. Combining the two online advertisement methods shows that indeed you can better measure the impact of the channels together as mentioned by Dinner et al (2014).
Hypothesis 2a ‘’There is a positive synergy effect of advertising between YouTube and website advertisement on offline purchases ‘’ was accepted looking at impact on sales, but rejected when looking at impact on purchase probability. When measuring the impact of the variables on sales, exposure to RTL advertisements turned out to have an insignificant correlation with Sales, but it is significantly positive correlated with YouTube exposure. Therefore, we can label this variable as a ‘’classical suppressor variable’’ (Baron and Kenny, 1986), (Ludlow et al, 2014), as RTL turns out to be statistically positively significant in combination with YouTube. The logistic regression showed a different result, as the synergy was insignificant. This means that advertising via YouTube and RTL only increases the sales for people that are already buying the product, but does not add to the explaining of the purchase probability. This is contradicting the work of Manchanda et al (2006) who found that online advertisement does affect purchase probabilities. Using more than one unique website to show the pre-roll advertisement like the one on the RTL website should increase the purchase probability (Manchanda et al, 2006).
Hypothesis 2b ‘’ There is a positive synergy effect of advertising between YouTube and TV campaigns on offline purchases’’ was accepted based on the impact on sales and rejected on the impact on purchase behavior. The positive synergy effect between YouTube and TV is consistent with the research of Danaher (2013) who wrote about the synergy achieved between TV advertisement and Social Media, of which YouTube is part of as a ‘content community’.
This suggests that the three advertisement tools of Brand X together do not create added value (1+1=3). Research by Chang and Thorson (2007) already pointed out that they achieved a synergy between online and offline media on purchase probability, something that was not the case in this study, where the results were insignificant. This shows again that within this dataset, the synergies only exist between two variables and when explaining sales, and not purchase probability.
Managerial implications
In the past, academics have shown that the marketing instruments affect the sales and purchase probability of products and that synergies can be achieved within an advertising campaign. Therefore, the advantages of marketing cannot be denied, as also shown in this study based on the results gathered from a dataset of one of the biggest FMCG companies in the world.
Synergies can be achieved, as shown in previous studies as well as in this study. Managers can achieve synergies by using what Fennis (2015) calls the IMC (Integrated Marketing Communications) program, which involves coordinating elements of the promotional mix to create synergy between them (Belch and Belch, 2004). To use a program like IMC, a company should take into account that an experienced manager should run the program and the program should be coordinated by one person to focus on the central message. Managers should be aware that the bigger the synergies they are trying to create, the bigger the budget needs to be (Naik et al, 2003). Overall, there are some insights for companies that want to achieve synergies found in different articles (Naik et al, 2003, Wang, 2006, Rice et al, 1998, De Pelsmacker et al, 2001):
- Send out a consistent message;
- Using the most relevant channels for the customers or prospects; - Examine the path of purchase to find the best marketing tools; - Coordinate the program to achieve synergies via one manager;
- Let the program be run via one marketing department with full control over all communications worldwide.
opposing another message send out, and is reinforcing it (Keller, 2015). This way, using multiple marketing tools is likely to boost sales. Brand X should also consider using more than one website to show banner advertisement, as shown in the literature. Brand X might also consider spending more money on YouTube advertisement, as this has been proven to be equally as good as TV advertisement. Brand X should also consider to not decrease its spending in RTL pre-roll advertisement as it is good to allocate more than fair share to the less effective medium, in this case RTL (Naik, Raman, 2009).
Limitations and further research
between two marketing tools is established in this study, giving directions to measure all three with another dataset.
Literature
Abraham, M. (2008). The off-line impact of online ads. Harvard Business Review, 86(4), 28. Assael, H. (2011). From silos to synergy. Journal of Advertising Research, 51(1 50th
Anniversary Supplement), 42-58.
Assael, H., & Poltrack, D. F. (1993). Using single source data to select TV programs. Journal of Advertising Research, 33(1), 48-57.
Baron, R. M., & Kenny, D. A. (1986). The moderator–mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of personality and social psychology, 51(6), 1173.
Belch, G.E. and Belch, M.A. (2004). Advertising and promotion management: An integrated marketing communications perspective (6th edn). Boston, MA: Irwin.
Briggs, R., Krishnan, R. & Borin, N. 2005. Integrated multichannel communication strategies: Evaluating the return on marketing objectives—the case of the 2004 Ford F-150 launch. Journal of Interactive Marketing, 19(3): 81–90.
Bruner, G. C., & Kumar, A. (2000). Web commercials and advertising hierarchy-of-effects. Journal of Advertising Research, 40(1-2), 35-42.
Campbell, C., Pitt, L. F., Parent, M., & Berthon, P. R. (2011). Understanding consumer conversations around ads in a Web 2.0 world. Journal of Advertising, 40(1), 87-102. Chang, Y., & Thorson, E. (2004). Television and web advertising synergies. Journal of Advertising, 33(2), 75-84.
intervals. Statistics in medicine, 28(2), 205-220.
Danaher, P. J., & Dagger, T. S. (2013). Comparing the relative effectiveness of advertising channels: A case study of a multimedia blitz campaign. Journal of Marketing Research, 50(4), 517-534.
Danaher, P. J., & Rossiter, J. R. (2011). Comparing perceptions of marketing communication channels. Journal of Marketing, 45(1/2), 6-42.
Dijkstra, M., Buijtels, H. E. J. J. M., & van Raaij, W. F. 2005. Separate and joint effects of medium type in consumer responses: A comparison of television, print, and the internet. Journal of Business Research, 58 (3): 377–386.
Dinner, I. M., Van Heerde, H. J., & Neslin, S. A. (2014). Driving online and offline sales: The cross-channel effects of traditional, online display, and paid search advertising. Journal of Marketing Research, 51(5), 527-545.
Drèze, X., & Hussherr, F. X. (2003). Internet advertising: Is anybody watching?. Journal of interactive marketing, 17(4), 8-23.
Edell, J. A., & Keller, K. L. (1989). The information processing of coordinated media campaigns. Journal of Marketing Research, 26(2), 149-163.
Fennis, B. M., & Stroebe, W. (2015). The psychology of advertising. Psychology Press.
Gibson, L. D. (1996). What can one TV exposure do?. Journal of Advertising Research, 36(2), 9-19.
Havlena, W., Cardarelli, R. & de Montigny, M. (2007) Quantifying the Isolated and
Hennig-Thurau, T., Malthouse, E. C., Friege, C., Gensler, S., Lobschat, L., Rangaswamy, A., & Skiera, B. (2010). The impact of new media on customer relationships. Journal of service research, 13(3), 311-330.
Kahneman, D. (1973). Attention and effort (p. 246). Englewood Cliffs, NJ: Prentice-Hall. Kaplan, A. M., & Haenlein, M. (2010). Users of the world, unite! The challenges and opportunities of Social Media. Business horizons, 53(1), 59-68.
Kumar, A., Bezawada, R., Rishika, R., Janakiraman, R., & Kannan, P. K. (2016). From Social to Sale: The Effects of Firm-Generated Content in Social Media on Customer Behavior. Journal of Marketing, 80(1), 7-25.
Lewis, R. A., & Reiley, D. H. (2014). Online ads and offline sales: measuring the effect of retail advertising via a controlled experiment on Yahoo!. Quantitative Marketing and Economics, 12(3), 235-266.
Lin, C., Venkataraman, S., & Jap, S. D. (2013). Media multiplexing behavior: Implications for targeting and media planning. Marketing Science, 32(2), 310-324.
Liu, Y., & Lopez, R. A. (2014). The impact of social media conversations on consumer brand choices. Marketing Letters, 1-13.
Loewenstein, J., Raghunathan, R., & Heath, C. (2011). The repetition-break plot structure makes effective television advertisements. Journal of Marketing, 75(5), 105-119.
Lodish, L. M., Abraham, M., Kalmenson, S., Livelsberger, J., Lubetkin, B., Richardson, B., & Stevens, M. E. (1995). How TV advertising works: A meta-analysis of 389 real world split cable TV advertising experiments. Journal of Marketing Research, 125-139.
Malhotra, N. K. (2010). Marketing research: An applied orientation (Vol. 834). New Jersey: Pearson Education.
Manchanda, P., Dubé, J. P., Goh, K. Y., & Chintagunta, P. K. (2006). The effect of banner advertising on internet purchasing. Journal of Marketing Research, 43(1), 98-108.
Moorman, M., Neijens, P. C., & Smit, E. G. (2007). The effects of program involvement on commercial exposure and recall in a naturalistic setting. Journal of Advertising, 36(1), 121- 137.
Naik, P. A., & Peters, K. (2009). A hierarchical marketing communications model of online and offline media synergies. Journal of Interactive Marketing, 23(4), 288-299.
Naik, P.A, & Raman, K. (2003). Understanding the impact of synergy in multimedia communications. Journal of Marketing Research, 40, 375-388.
Ngai, E. W. (2003). Internet marketing research (1987-2000): a literature review and classification. Journal of Marketing, 37(1/2), 24-49.
Pavlou, P. A., & Stewart, D. W. (2000). Measuring the effects and effectiveness of interactive advertising: A research agenda. Journal of Interactive Advertising, 1(1), 61-77.
Rangaswamy, A., & Van Bruggen, G. H. (2005). Opportunities and challenges in multichannel marketing: An introduction to the special issue. Journal of Interactive Marketing, 19(2), 5-11.
Rigby, D. (2011). The future of shopping. Harvard Business Review, 89(12), 65-76.
Romaniuk, J., Beal, V., Uncles, M. (2013), Achieving Reach in a Multi-Media Environment. Journal of Advertising Research, 53(2): 221-230.
Schramm-Klein, H., Wagner, G., Steinmann, S., & Morschett, D. (2011). Cross-channel integration–is it valued by customers?. The International Review of Retail, Distribution and Consumer Research, 21(5), 501-511.
Sethuraman, R., Tellis, G. J., & Briesch, R. A. (2011). How well does advertising work? Generalizations from meta-analysis of brand advertising elasticities. Journal of Marketing Research, 48(3), 457-471.
Sloot, L. (2016). Retail Masterclass of the EFMI Business School for students of the Rijksuniversiteit Groningen and University of Amsterdam.
Spilker-Attig, A., & Brettel, M. (2010). Effectiveness of online advertising channels: A price-level-dependent analysis. Journal of Marketing Management, 26(3-4), 343-360.
Stammerjohann, C., Wood, C.M. Chang,Y. & Thorson, E. 2005. An Empirical Investigation of the Interaction between Publicity, Advertising, and previous Brand Attitudes and
Knowledge. Journal of Advertising, 34 (4): 55–67.
Tellis, G. J., & Weiss, D. L. (1995). Does TV advertising really affect sales? The role of measures, models, and data aggregation. Journal of Advertising, 24(3), 1-12.
Verhoef, P. C., Kannan, P. K., & Inman, J. J. (2015). From multi-channel retailing to omni-channel retailing: Introduction to the special issue on multi-omni-channel retailing. Journal of Retailing, 91(2), 174-181.
Verhoef, P. C., & Leeflang, P. S. (2009). Understanding the marketing department's influence within the firm. Journal of marketing, 73(2), 14-37.
Wagner, G., Schramm-Klein, H., & Steinmann, S. (2013). Effects of cross-channel synergies and complementarity in a multichannel e-commerce system–an investigation of the
Wakolbinger, L. M., Denk, M., & Oberecker, K. (2009). The Effectiveness of Combining Online and Print Advertisements Is the Whole Better than the Individual Parts? Journal of Advertising Research, 49(3), 360-372. https://www.YouTube.com/yt/press/statistics.html http://www.forbes.com/sites#/sites/theyec/2014/05/08/looking-for-new-ways-to-reach-customers-try-YouTube-advertising/#4981d5057add (Siu, 2014) http://www.emarketer.com/Article/UK-Achieve-World-First-Half-of-Media-Ad-Spend-Goes-Digital/1012280 Appendix 1.1 Multicollinearity Predictors Unstandardized Coefficients t Sig. Collinearity Statistics
B Std. Error Beta Tolerance VIF
