What Matters Most in Advertising Campaigns

download What Matters Most in Advertising Campaigns

of 23

description

adveritisng

Transcript of What Matters Most in Advertising Campaigns

  • What matters most inadvertising campaigns?

    The relative effect of media expenditureand message content strategy

    Bas van den PutteUniversity of Amsterdam

    Three main factors determine the effect of advertising campaigns: message content strat-egy, advertising expenditure and previous consumer behaviour. This study investigatesthe relative strength of each of these influences. Four possible campaign targets are takeninto account: campaign recall, campaign likeability, brand recall and intention to purchasethe advertised brand. A study of 40 advertising campaigns, evaluated on average by 1200consumers, shows that, after controlling for the effect of previous purchase behaviour,the effect of message content strategy is generally larger than the effect of advertisingexpenditure. Nevertheless, more media expenditure positively influences campaignrecall and influences brand recall and purchase intention of small brands. The most effec-tive message content strategy differs per campaign target, but overall awareness and like-ability strategies are most effective. For purchase intention only, the emotions strategycan be advised. The information strategy is best avoided.

    IntroductionEvery year, more and more money is spent on advertising. Between 1997and 2006, television expenditures rose by 24% in the United Kingdom,by 43% in the Netherlands, and as much as 57% in the US (Trends inAdvertising Expenditure 2007). But is all this money well spent.? Lodishet al. (1995) found that increasing the television advertising budget waseffective for 55% of the brand introductions studied, but only for 33% ofthe established brands (though a recent update of this study showed thatadvertising campaigns became more effective after 1995 - Hu et al. 2007).In a test of 63 on-air commercials, Gibson (1996) found that only 19% had

    International Journal of Advertising, 28(4), pp. 669-690 2009 Advertising AssociationPublished by the World Advertising Research Center, www.warc.comDOI: 10.2501/S0265048709200813 669

  • INTERNATIONAL JOURNAL OF ADVERTISING, 2009, 28(4)

    a significant positive effect on brand preference of consumers who saw thecommercial (and 8% of the commercials even had a significant negativeeffect). Jones (1995a) studied the campaigns for 78 brands and found that70% of them had a positive short-term effect, but only 46% had a positivelong-term effect. Based on these and similar studies of real-world advertis-ing, both Maclnnis et al (2002) and Tellis et al (2005) concluded that, onaverage, increased advertising exposure had only small positive effects onthe responses of consumers, if any.

    These results raise the question of how advertising budgets can be bet-ter spent in order to stimulate more consumers to purchase the advertisedbrands. Of course, the amount of advertising is not the only determinantof advertising success. The content of the advertisement can also matter:the message content strategy. A huge advertising budget, making surethat the target group encounters the advertising campaign several times,is useless if the commercials themselves do not convey an effective mes-sage. Likewise, a brilliantly made commercial is useless if no one sees it.Both advertising expenditure and message content strategy are under theadvertiser's control, though limited by practical and financial considera-tions, and each has an important influence on the advertising effects.

    The effects of media spending and message content variables havebeen studied extensively. However, few studies have compared the effectof message content strategies with the effect of advertising expenditure,though there is some evidence from an early study that message qualityhad far more influence than advertising expenditure on changes in mar-ket share (Buzzell 1964). Aaker and Carman (1982) gave an overview ofstudies that tested either advertising expenditure or advertising copy, butnone of these tested expenditure and copy simultaneously. They foundthat increased advertising expenditure had a positive effect on short-termsales in 30% of the studies, whereas variation in advertising copy showedsignificant differences in 47% of the studies. These results suggest thatvariation in message content strategy has more effect than variation inadvertising expenditure. However, it is hard to draw a conclusion, becausenone of the studies examined the effect of variation in message strategyand advertising expenditure simultaneously. The present study combinedboth variables in order to gain more insight into the relative influence ofmedia spending and message content strategies on advertising effect vari-ables. Because effects might not be independent, the interaction effect

    670

  • WHAT MATTERS MOST IN ADVERTISING CAMPAIGNS?

    between message content strategies and media expenditure was alsotaken into account.

    Of course, there are other relevant independent variables in additionto media expenditure and message content strategies, the most importantof which is probably past purchase behaviour. It is likely that consumersremember the brands they usually buy the best, and that they will buy thesame brands again. Many people are loyal to one brand or to a small set ofbrands. For example, Batra etal. (1995) found that advertising expenditurehad no effect on purchase intention ( = 0.01), but that previous behaviourhad a large effect on it ( = 0.79). Tellis (1988) found similar results.

    The present study tested a simple model to explain consumer behav-iour that included all three of these aspects: media expenditure, messagecontent strategies and previous consumer behaviour. The study examinedwhich of these three variables had the most influence on campaign recall,campaign appreciation, brand recall and purchase intention. This studywas unique in several aspects. First, these three factors have never beencombined in one research design before. Whereas some studies haveexamined the effect of media expenditure (e.g. Jones 1995a; McDonald1996), and others have investigated the effect of message content ele-ments (e.g. Stewart & Furse 1986; Belch et al. 1987), media expenditureand content have seldom been combined in one study. Second, thoughmany studies have shown that past behaviour has an enormous influenceon future behaviour (e.g. Tellis 1988; Batra et al. 1995), the influence ofprevious behaviour on advertising effects has often been ignored. Third,whereas previous studies on content elements have often employed forcedexposure in theatre settings or experiments (e.g. Stewart & Furse 1986;Laskey et al. 1995), the present study investigated the effects of naturalexposure to campaigns, by interviewing consumers who were exposed tothe campaign over several weeks in their own homes during the periodthat the campaign aired on television.

    Message content strategiesThere are many ways to look at message content strategies, and varioustypologies have been developed (e.g. Rossiter et al. 1991; Laskey et al.1995; Taylor 1999). The present study used a typology originally devel-oped by Franzen (1999) and further developed by Van den Putte (2002,

    671

  • INTERNATIONAL JOURNAL OF ADVERTISING, 2009, 28(4)

    2006). This typology stipulates that every successful brand must meetthree basic conditions.

    First, people must know the brand, be aware of it. Awareness is impor-tant because consumers generally do not choose brands that are unknownto them (Rossiter etal. 1991). They tend to prefer brands that are knownto them, even if they are confronted with unfamiliar brands of better qual-ity (e.g. Hoyer & Brown 1990; MacDonald & Sharp 2000). Brand aware-ness can be achieved by designing a campaign that stands out, by doingsomething different from the competition, by doing something that drawsattention. This is called the awareness strategy.

    Second, people must like the brand. This goal can be achieved by usinghumour or by amusing people in another way. This is called the likeabilitystrategy. The rationale for this strategy is that appreciation of the advertise-ment is transferred to the brand, which is thus also liked (Muehling &MeCann 1993).

    Third, the brand must fulfil a consumer need. This need may alreadybe felt by the consumer, or it can be induced by the campaign. Whereasthe awareness and likeability strategy are more or less relevant for allproduct categories - though the creative execution can differ - the type ofeonsumer need varies enormously over product categories or even brands.The message content strategy thus depends on the relevant consumerneed: it can be instrumental, emotional, hedonistic, normative or social(for the corresponding strategies, see Van den Putte 2002; Van den Putte& Dhondt 2005). The present study concentrated on two major strategies,the information strategy and the emotions strategy. These appear in manytypologies and are the ones most relevant to the product categories used inthis study: cleaning products and food products. The information strategycommunicates instrumental advantages. For example, people buy wash-ing powder primarily to make their laundry clean, so a commercial mightgive information about the instrumental advantages of the product. Theemotions strategy eoncentrates on the emotional consequences of productuse, creating a completely different kind of commercial - for example, ahappy family enjoying a cup of coffee. Finally, another strategy is also rele-vant for this study, the sales-response strategy. Campaigns using this strategymight mention a temporary discount, or offer consumers a gift if they buythe brand this week. This is an important strategy because research hasshown that advertising in combination with promotions had more effect

    672

  • WHAT MATTERS MOST IN ADVERTISING CAMPAIGNS.?

    than advertising that was not combined with promotions (Jones 1995b).Bemmaor and Mouchoux (1991) found that advertised price reductionsincreased sales between 20% and 180% more than unadvertised in-storeprice reductions.

    iVIedia expenditureMedia expenditure is often evaluated using GRPs (gross rating points).This is a statistic that indicates the reach of a campaign among members ofthe target group. The GRP measure takes into account both the number ofpeople that are exposed and the number of times each person is exposed.It is generally assumed that the more GRPs a campaign achieves, themore likely it is to achieve its communication and marketing objectives.Of course, the success of a campaign may depend not only on the GRPsit achieves, but also on the GRPs that competitors' campaigns achieve.Perhaps, it is less important how much a brand is advertised than that it beadvertised more than other brands in the same product category. Thus, adistinction must be made between absolute and relative GRPs. The rela-tive GRPs represent the GRPs achieved as a percentage of total numberof GRPs obtained by all competing brands in the product category. This isgenerally called share-of-voice. The absolute number of GRPs, independ-ent of competitors' efforts, is called voice in this study.

    Campaign effectsSo far, the term campaign effects has been used in a general way. Of course,this is much too simple, as a campaign can have many different effects.The first distinction that needs to be made is between what people think,feel and know about the commercials in the campaign - the advertisementeffects - and what people think, feel and know about the advertised brand- the brand effects. This study considers two advertisement effects - cam-paign recall and campaign appreciation - and two brand effects - brandrecall and purchase intention.

    The awareness strategy is the best strategy to increase campaign recalland brand recall because it attempts to attract attention to the campaignand brand. Gampaign and brand recall are also influenced by voice andshare-of-voice. Most likely, share-of-voice has more effect than voice.

    673

  • INTERNATIONAL JOURNAL OF ADVERTISING, 2009, 28(4)

    because campaigns that achieve more exposure relative to their com-petitors are more successful in increasing recall (Rossiter & Percy 1998,pp. 222-223). If brand recall is already large, it might be difficult toincrease it further. Therefore, larger effects are expected for small brands.

    The best way to increase campaign appreciation is the use of the like-ability strategy. Because appreciation of a campaign depends mainly onthe campaign's content, it will not be influenced by media expenditure.Nevertheless, increase in voice might have a negative effect on campaignappreciation if consumers get irritated as a result of overexposure.

    To influence purchase intention, the campaign should address relevantconsumer needs. Therefore, the information strategy is probably mosteffective for cleaning products, and the emotions strategy for food prod-ucts. Eor both product categories, it is also important that people are awareof the brand and have a generally positive feeling towards it. Therefore,the awareness and likeability strategy might be effective for both productcategories. This equally holds for the sales-response strategy because atemporary discount probably has a positive effect on purchase intention.Intention to purchase a brand will also be increased by media expenditure,especially share-of-voice. It is expected that this effect will be stronger forsmall brands because there are more consumers that have not yet triedthese brands.

    Overview of the studyThis study examined the effects of 40 advertising campaigns in two mainproduct categories: cleaning products and food products. Four effectswere included: campaign recall, campaign appreciation, brand recall andpurchase intention. Three types of influence were considered: messagecontent strategies, media expenditure and previous consumer behaviour.The main aim of this study was to see which of these three independentvariables influenced the four dependent variables most.

    MethodAs mentioned, 40 advertising campaigns were studied: 18 campaigns forcleaning products and 22 campaigns for food and candy products. Thesetwo main product categories were further divided into ten subcategories

    674

  • WHAT MATTERS MOST IN ADVERTISING CAMPAIGNS?

    (e.g. all-purpose cleaner, dishwasher detergent, meal sauces, chocolatebars), so that share-of-voice could be calculated. This was calculated asa percentage of the total GRPs achieved by all competing brands in asubcategory.

    Intomart GfK, a Dutch market research organisation, conducted theinterviews on behalf of SPOT (Foundation for Promoting and OptimisingTelevision Advertisements). A random sample of Dutch households wastaken, and telephone interviews were conducted with the person in eachhousehold who did most of the daily shopping. The response rate was30%. Each respondent answered questions about two of the 40 campaigns.On average, 1200 respondents were interviewed per campaign. The sur-vey was conducted over a period of 30 weeks, but respondents were askedonly about campaigns that had run on television in the eight weeks priorto their interview.

    Intomart GfK also maintains 2L people meter panel o 1300 households thatis used by Dutch advertising companies to measure the GRPs a campaignachieves on national television. The data from this panel were used to esti-mate the GRPs of each of the campaigns in the eight weeks prior to eachinterview. Both voice (absolute number of GRPs that a campaign reached)and share-of-voice (GRPs reached compared to the GRPs reached by allcompeting brands in the same product category) were estimated. Thoughboth the people meter panel and the survey sample were random samplesof the Dutch population, sample differences were taken into account. Thenumber of GRPs found in the complete people meter panel was correctedby taking into account the characteristics of the respondents in the surveysample (including their television viewing behaviour) in order to estimateas closely as possible the number of GRPs achieved in the survey sample.This varied over the campaigns from six GRPs to 954 GRPs.

    QuestionnaireThis study focused on differences between campaigns. Thus, the advertis-ing campaign was the unit of analysis, rather than the individual respond-ent. The average scores for campaign recall, campaign appreciation, brandrecall, purchase intention and previous behaviour were calculated over allrespondents. The following questions were asked (in the order presentedhere).

    675

  • INTERNATIONAL JOURNAL OE ADVERTISING, 2009, 28(4)

    Brand recall: an open question asked respondents to name all the brandsthey could think of in a certain product category (e.g. all-purpose clean-ers). Brand recall was operationalised as top-of-mind brand recall - thatis, the flrst brand that a respondent mentioned. On average, the brandsin this study were top-of-mind for 13% of the respondents, rangingbetween 1% and 65%.

    Campaign recall: respondents indicated which advertisements theyremembered having seen in the past few weeks for a certain productcategory. In order to correct for false memory, respondents were askedto describe the content of the commercial to check whether they hadseen a recent commercial. A positive recall was coded only if respond-ents correctly remembered at least one content element of the recentcommercial. On average, 11% of the respondents (ranging between0.02% and 33%) could spontaneously recall a speciflc campaign.

    Campaign appreciation: respondents gave each campaign they couldremember a score between 1 and 10. On average, the campaigns scoreda 6.5, varying between 4.6 and 8.5. In order to measure campaignappreciation, brands that respondents had not recalled at the previousquestion were mentioned, and if respondents correctly remembered thecampaign after this cue, their appreciation was measured.

    Previous behaviour, respondents were flrst asked what brand they usu-ally buy, and then what other brands they occasionally buy. Previousbehaviour was deflned by combining both items (i.e. the set of brandsthat were either usually bought or occasionally bought). This deflnitioncorrelated .97 with a deflnition of previous behaviour that includedonly the most often bought brand. Brands were usually or occasionallybought by between 2% and 73% of the respondents.

    Purchase intention: for each product category, respondents were askedwhat brand they planned to buy at their next purchase. On average,10% of the respondents intended to buy a particular brand, rangingbetween 1% and 47%.

    Content analysis

    Using the description of the message content strategies formulated above(based on Franzen 1999), a coding form was developed. Three trained

    676

  • WHAT MATTERS MOST IN ADVERTISING CAMPAIGNS.?

    coders independently coded the message content strategies used in eachcampaign. A code was assigned only if the three coders were unanimous.If they disagreed on the assignment of a particular strategy to a campaign,the code for this strategy was set to missing. Overall, the agreement amongthe coders was extremely high. In fact, they disagreed only twice, oncewith respect to the use of the awareness strategy and once with respect tothe emotions strategy.

    Most campaigns consisted of a number of commercials (between oneand six), which almost always used the same strategies. Of course, a cam-paign can use several message content strategies - for example, combiningthe information strategy with the likeability strategy. If a certain strategywas used in at least one commercial in the campaign, this strategy wascoded as being used in the campaign.

    In the coding form, the following yes-no questions were used to codethe message content strategies for each commercial in each of the 40campaigns.

    Awareness strategy: 'Does the commercial differ from what is normalfor the product category (e.g. in content or style).?' and 'Does the com-mercial use animation or special effects.?'

    Likeability strategy: 'Is the commercial intended to be humorous.?','Does the commercial aim to amuse viewers (other than by beinghumorous).?' and 'Are there any "cuties" in the commercial, such as petsor children.?'

    Emotions strategy: 'Does the commercial aim to touch the feelings ofthe consumer or does the commereial show what feelings can be expe-rienced by using the brand.?'

    Information strategy: 'Does the commercial give information about(new) instrumental advantages, qualities, or properties of the brand.?' or'Is the commercial primarily rational (e.g. by giving instrumental advan-tages or disadvantages).?'

    Sales-response strategy: 'Does the commercial offer a temporary dis-count.?' or 'Does the commercial include other forms of sales promotion,such as gifts or prize contests.?'

    677

  • INTERNATIONAL JOURNAL OF ADVERTISING, 2009, 28(4)

    ProcedureBivariate correlations, means and standard variation of all variables aregiven in Table 1. Regression analyses were used to test the relationshipsbetween the four advertising effect variables and message content strate-gies, media expenditure and previous purchase behaviour. Because theadvertising campaign was the unit of analysis in this study, 40 cases wereavailable. This limited the number of independent variables that could beincluded in each regression analysis, so composite message content vari-ables were constructed instead of entering each of the five coded adver-tising strategies separately. Because different message content strategieswere effective for each campaign effect, four composite message contentvariables were constructed, one for each campaign effect (i.e. campaignrecall, campaign appreciation, brand recall and purchase intention).

    In order to find the most effective advertising strategies, each cam-paign effect was regressed on message content strategy, previous purchasebehaviour and product category (0 = cleaning product, 1 = food or candyproduct). Because the effect of content strategy might differ by product

    Table 1: Bivariate

    Variable

    1. Purchase intention

    2. Brand recall

    3. Campaign appreciation

    4. Campaign recall

    5. Awareness strategy

    6. Likeability strategy

    7. Information strategy

    8. Emotions strategy

    9. Sales-responsestrategy

    10. Voice

    11. Share-of-voice

    12. Past purchasebehaviour

    M

    SO

    correlations, means and standard deviations

    1

    1.00

    0.97

    0.00

    0.28

    -0.03

    -0.22

    0.18

    -0.14

    0.05

    0.35

    0.37

    0.98

    10.56

    10.84

    2

    1.00

    -0.04

    0.25

    0.05

    -0.18

    0.14

    -0.06

    0.07

    0.31

    0.30

    0.96

    13.67

    16.14

    3

    1.00

    0.36

    0.29

    0.26

    -0.36

    0.26'

    -0.07

    -O.09

    0.30

    0.02

    6.48

    0.67

    4

    1.00

    0.27

    0.25

    -0.14

    0.09

    -0.25

    0.10

    0.42

    0.31

    10.79

    8.83

    5

    1.00

    0.28

    -0.41

    0.48

    0.21

    -0.07

    0.06

    -^).O4

    0.64

    0.49

    6

    1.00

    -0.35

    0.39

    -0.10

    -0.04

    0.03

    -0.21

    0.80

    0.41

    7

    1.00

    -0.95

    0.17

    0.15

    -0.28

    0.16

    0.53

    0.51

    8

    1.00

    -0.21

    -0.16

    0.30

    -0.12

    0.51

    0.51

    9

    1.00

    0.20

    -0.19

    0.01

    0.18

    0.38

    10

    1.00

    0.53

    0.31

    247.27

    198.52

    11 12

    1.00

    0.32 1.00

    0.63 20.30

    0.23 16.66

    678

  • WHAT MATTERS MOST IN ADVERTISING CAMPAIGNS?

    category, an interaction effect between both variables was added. Aninteraction effect between previous purchase behaviour and content strat-egy was also added because the effect of advertising strategies has beenshown to depend on the amount of experience with a brand (Van denPutte 2006). This regression analysis was done separately for each contentstrategy. Content strategies that had a significant regression coefficient, atp

  • INTERNATIONAL JOURNAL OF ADVERTISING, 2009, 28(4)

    but no problems were found. For each analysis, standardised residualsand Cook's distance were checked in order to find outliers and influentialcases. Two campaigns were removed from some analyses because theywere problematic. One of these campaigns might have been an outlierbecause it was for a product introduction. Only 2% of the respondents hadpreviously bought the advertised product, which was the lowest figure ofall brands. The other problematic campaign was for an established brand,which was occasionally bought by 35% of the consumers, though only 5%intended to purchase it next time. Of all campaigns, this showed the larg-est difference between purchase intention and previous behaviour.

    Results

    Construction of the composite message content strategyvariablesFirst, the most effective content strategies were estimated separately foreach campaign effect (Table 2). Campaign recall was largest when eitherthe likeability strategy or the awareness strategy was used. An interactioneffect between product category and awareness strategy indicated that thisstrategy was twice as effective for food products than for cleaning prod-ucts. The information strategy had a positive effect on campaign recall forcleaning products. Campaign appreciation was found to be largest withthe awareness strategy and likeability strategy. Campaign appreciationwas smaller when the information strategy was used. The only effectivecontent strategy for brand recall was the awareness strategy. The effectsof content strategies on purchase intention depended on brand size.Interaction effects between previous purchase behaviour and messagestrategy indicated that when fewer consumers regularly or occasionallybought a brand, the likeability and emotions strategies were more effec-tive, whereas the information strategy was less effective. Purchase inten-tion was somewhat larger for cleaning products than for food and candyproducts, but though an interaction effect between product category andcontent strategy was expected, this was not found. Also, no effect of thesales-response strategy was found.

    680

  • WHAT MATTERS MOST IN ADVERTISING CAMPAIGNS.?

    Table 2: Regression of the four campaign effects on previous purchase behaviour,product category, and each of the five message content strategies

    Awareness strategyPrevious purchase behaviourProdua categoryAwareness * Previous behaviourAwareness * Product categoryLikeability strategyPrevious purchase behaviourProduct categoryLii

  • INTERNATIONAL JOURNAL OF ADVERTISING, 2009, 28(4)

    Table 3: Regression of the four campaign effects on previous behaviour, voice,message content strategy and product category

    Campaign Campaign Purchaserecall appreciation Brand recall intention

    (fi = 39) (n = 39) (n = 38) (n = 37)

    Voice (absolute GRPs)Message content strategy

    Voice * Message strategy

    Previous purchase behaviour

    Product category

    Note: For campaign recall, adjusted R' = 0.21 (p < 0.05). for campaign appreciation, adjusted fi^ = 0.15 (p < 0.10). For brandrecali, adjusted R' = 0.93 (p < 0.001). For purchase intention, adjusted R' = 0.97 (p < 0.001).p < 0.10; *p < 0.05; * *p < 0.01; * * *p < 0.001

    0.12

    0.46**

    0.05

    0.40*

    0.09

    -0.14

    0.40*

    -0.29

    0.17

    0.09

    0.03

    0.10*

    0.05

    0.94***

    -0.02

    0.03

    0.09**

    -0.03

    0.99***

    -0.06

    Strategy variable and previous purchase behaviour. This indicated that useof the most effective message content strategies had more influence oncampaign recall than increasing the absolute number of GRPs. Messagecontent strategy also had the greatest influence on campaign appreeiation,but as absolute number of GRPs grew, the effect of the message contentstrategy tended to decrease. Both brand recall and purchase intentionwere almost entirely influenced by previous purchase behaviour, but forboth of these variables a small but signiflcant effect of message contentstrategies was also found. Purchase intention was slightly smaller for foodproducts. No direct effect was found for absolute number of GRPs for anyof the four campaign effects.

    Table 4 shows the results of the regression analyses testing the effect ofshare-of-voice. Gampaign recall was most strongly influenced by messagecontent strategy and by share-of-voice. Gampaign appreciation was mostStrongly influenced by message content strategy, though the effect of con-tent strategy became smaller when share-of-voice increased. Brand recalland purchase intention were almost entirely influenced by previous pur-chase behaviour. In addition, message content strategy had a significanteffect on both brand recall and purchase intention, whereas share-of-voicehad no effect.

    A comparison of the regression coefficients of voice and share-of-voice(Tables 3 and 4) showed that share-of-voice had more influence thanvoice on the advertisement effects, in particular campaign recall. For

    682

  • WHAT MATTERS MOST IN ADVERTISING CAMPAIGNS.?

    Table 4: Regression of the four campaign effects on previous behaviour, share-of-voice, message content strategy and product category

    Share-of-voice (SOV; relative GRPs)Message content strategy

    SOV * Message strategy

    Previous purchase behaviour

    Produrt category

    Campaignrecall

    (n = 39)0.37*

    0.36*

    0.08

    0.26

    -0.08

    Campaignappreciation

    (n = 39)0.21

    0.38*

    -O.28#

    0.08

    0.11

    Brand recall(n = 38)-0.01

    0.10*

    0.06

    0.95***

    -0.04

    Purchaseintention(n = 37)0.00

    0.10**

    -0.03

    0.99***

    -0.07*

    Note: For campaign recall, adjusted fi' = 0.32 (p < 0.01). For campaign appreciation, adjusted R' = 0.21 (p < 0.05). For brandrecall, adjusted R' = 0.93 (p < 0.001). For purchase intention, adjusted R' = 0.97 (p < 0.001).p < 0.10; *p < 0.05; * *p < 0.01; * * *p < 0.001

    brand effects, neither voice nor share-of-voice influenced brand recall andpurchase intention. As was suggested above, brand recall is difficult toincrease when it is already large, and more effect was expected for brandsthat scored low on brand recall. Therefore, brands with a brand recallbelow 10% were compared with those with a recall of at least 10%. Table5 shows that both voice and share-of-voice had more positive effect on thelow-recall brands. Message content strategy also had more positive effectwhen brand recall was low. Still, these results must be considered tenta-tive for two reasons. First, brand recall before exposure was unknown.Therefore, brand recall after exposure to the campaign was used tosplit the campaigns into high- and low-recall groups. Nevertheless, it isunlikely that the results would differ significantly if brand recall beforeexposure was known, because the level of brand recall was mainly influ-enced by previous purchase behaviour rather than by campaign exposure.Second, because the number of campaigns in each group was too small fora regression analysis with four independent variables, the results that arepresented in Table 5 might not be stable.

    A similar analysis was done for purchase intention. Previous purchasebehaviour was used to split the campaigns into two groups: small brandsand large ones. Brands that were bought by less than 15% of the respond-ents were defined as small brands, whereas brands that were bought by atleast 15% were defined as large brands. Table 5 shows that voice, share-

    683

  • INTERNATIONAL JOURNAL OF ADVERTISING, 2009, 28(4)

    Table 5: Regression of brand recall and purchase intentionbehaviour, media presence, message content strategy andcomparison between brands

    Voice (absolute GRPs)iVIessage content strategyPrevious purchase behaviourProduct categoryAdjusted R^Share-of-voice (reiative GRPs)Message content strategyPrevious purchase behaviourProduct category

    Adjusted R^

    p < 0,10; *p < 0,05; **p < 0,01; ***p < 0,001

    Brand recall

    on previousproduct category:

    Purchase intention

    Low recall High recall Small brands Large brands

    (" =0,260,320,83

    -0,200,480,260,250,79

    -0,44

    0,45

    21) (n=17)-0,07

    0,09* * * 0,94***

    -0,06** 0,88***

    -0,18#0,16#

    ** 0,96*** -0,08

    ** 0,91***

    (n = 20) (n = 16)0,12 0,010,17 0,030,88*** 0,95***

    -0,33* -0,080,73*** 0,96***0,29* -0,110,12 -0,030,89*** 0,99***

    -0 ,51* * -0,13*

    0,78*** 0,97***

    of-voice and message content strategy all had more effect on purchaseintention for small brands.

    Conclusion and discussionThis study investigates the influence of media expenditure, messagecontent strategies and previous behaviour on four campaign effects: cam-paign recall, campaign appreciation, brand recall and purchase intention.Most of the expectations formulated above are confirmed, though thereare some additional findings. Previous behaviour has a strong influence,especially on brand effects. Brand recall and purchase intention are almostentirely explained by previous purchase behaviour. These results are inline with Ehrenberg's view of advertising as a weak force whose mainaim is to maintain existing buyers (e.g. Ehrenberg 1974; Ehrenberg etal. 1997). Because of consumers' habitual buying behaviour, advertisingcampaigns can effect only small changes in purchase behaviour. Howeversmall, these effects can still have significant consequences. Suppose thatan advertising campaign influences the behaviour of just 1% of all con-sumers: this would make an enormous difference in sales. The results of

    684

  • WHAT MATTERS MOST IN ADVERTISING CAMPAIGNS.?

    this Study also indicate that it is important to include previous behaviourin studies of advertising effects. For example, if purchase intention hadbeen regressed on the message content strategies without taking previ-ous purchase behaviour into account, the content strategy of the largestbrand would always have appeared as the most effective, irrespective ofthe strategy employed.

    Advertisers control two main factors that determine the effect of adver-tising campaigns: media expenditure (i.e. GRPs) and message contentstrategy. Advertisement effects (i.e. campaign recall and campaign appre-ciation) as well as brand effects (i.e. brand recall and purchase intention)are more strongly related to the choice of message content strategy thanto media expenditure, with one exception: campaign recall is equallyinfluenced by message content strategy and share-of-voice (i.e. relativenumber of GRPs). However, voice (i.e. absolute number of GRPs) has nosignificant effect on campaign recall. This makes sense, because televisioncampaigns that are seen more often than other campaigns are more likelyto be remembered than those that are seen less. On the other hand, bothvoice and share-of-voice have a negative interaction effect with messagecontent strategy on campaign appreciation. This suggests that consumersdo not appreciate campaigns to which they are exposed too often. Turningto the brand effects, voice and share-of-voice have more positive influ-ence on brand recall for smaller brands than for larger ones. This couldbe because small brands are less well known and therefore have more togain from exposure. Because more consumers are familiar with the largerbrands, increasing brand recall will be much more difficult. Finally, share-of-voice has a positive effect on purchase intention, but only for smallbrands.

    The effects found for media expenditure were generally small, whichis somewhat surprising. The main reason is probably that this study con-trolled for the influence of previous purchase behaviour. When, in anadditional series of analyses, previous behaviour was omitted from theregression equation, voice ( = 0.27) and especially share-of-voice ( =0.54) were found to have large effects on purchase intention. While thissuggests that media expenditure has a large effect on overall sales, thiswould be a false conclusion. It is more likely that these effects are largelyexplained by the fact that (1) media expenditure is related to brand size(previous purchase behaviour has moderate correlations with voice, r =

    685

  • INTERNATIONAL JOURNAL OF ADVERTISING, 2009, 28(4)

    0.31, and share-of-voice, r = 0.32), and (2) most people purchase one ofthe brands that they normally buy (the correlation between purchaseintention and past behaviour is 0.98). Proper conclusions on the effect ofmedia expenditure can therefore be drawn only when previous behaviouris included in the analyses.

    Limitations and future researchA first limitation of this study is that it does not explain the effects of cam-paign exposure on loyal brand behaviour, because the effect of previouspurchase behaviour is controlled for in the analyses. The results show towhat extent GRPs and content strategy can explain why people intend topurchase a brand they do not usually or occasionally buy. Many advertis-ing campaigns are aimed in large part at preventing loyal consumers fromswitching brands or at encouraging them to purchase more of the brand.This study does not address these aims, concentrating instead on changesin brand preference.

    A second limitation of this study is that only a few product categoriesare analysed - namely, cleaning products, food and candy. It is unclearwhether similar results would hold for all product categories, especiallyfor products other than fast-moving consumer goods. Future researchshould include other product categories, and ideally each product categoryshould be analysed separately. However, the cost of collecting a sufficientnumber of cases per product category would be high, because hundreds ofconsumers need to be interviewed for each campaign.

    A third limitation of this study is that measures of message contentstrategy and media expenditure are relatively crude. The effects of mediaexpenditure also depend on variables that are not included in this study,such as the use of other media than television, the media schedule, andother marketing efforts (e.g. in-store promotion). Also, the five contentstrategies are not separately tested because the number of cases in thisstudy limits the number of variables that could be included in eachregression analysis. Instead, composite message content strategy variablesare calculated. In future studies incorporating more campaigns, it wouldbe interesting to see extended analyses that test each content strategyseparately.

    686

  • WHAT MAIT'ERS MOST IN ADVERTISING CAMPAIGNS.?

    ImplicationsThis study shows that, in so far as campaigns can influence consumers,choosing the potentially most effective message content strategy is oneof the most important decisions for the advertiser. The message contentstrategy should be chosen according to the campaign goal, and differ-ent market situations call for different message content strategies. Theawareness and likeability strategies are good choices for advertisers whowant to achieve advertisement effects (i.e. campaign recall and campaignappreciation). In addition, the information strategy has a positive effect oncampaign recall, but only for cleaning products, probably because peoplemore easily remember information that meets their expectations.

    For new brands, the first priority might be increasing brand recall. This isbest achieved by using the awareness strategy. Increasing purchase inten-tion is another important potential objective. This study suggests that thisis best achieved by using the emotions strategy in combination with thelikeability strategy, whereas the information strategy should be avoided.These results hold only for small brands because the purehase inten-tion of large brands is almost entirely determined by previous purchasebehaviour. The expected interaction effect between product categoryand content strategy on purchase intention is not found. The emotionsstrategy turned out to be more effective than the information strategyfor food and candy products as well as cleaning products. This is in linewith recent results of Binet and Field (2007), who report that the emo-tions strategy more effectively than the information strategy influencesbrand commitment and sales, even for rational products. The likeabilityand emotions strategies might have a more positive effect on purchaseintention for both cleaning and food products because most respondentsprobably have a low level of involvement with these products, particularlywhen they have less personal experience with the brand, as in the case ofthe small brands. When people have low levels of involvement, it is lesslikely that they will process the message eontent, particularly when theyare in a positive mood (Batra & Stayman 1990; Zhang & Zinkhan 2006).More likely, evaluation of the brand is influenced by a general evaluationof the advertisement, often based on peripheral cues, sueh as the use ofhumour in the likeability strategy, and this will have more effect than theinformational content of the message. As was seen, the likeability strategy

    687

  • INTERNATIONAL JOURNAL OF ADVERTISING, 2009, 28(4)

    makes respondents evaluate the advertisement more positively, whereasthe information strategy has a negative effect. In addition, when peoplewatch a commercial they like, their mood might become more positive,thus further decreasing the processing of information.

    Another conclusion of this study is that spending more money thancompetitors does help to increase campaign recall. But, if too much moneyis spent, campaign appreciation suffers because of excessive campaignexposure. Furthermore, increasing voice and share-of-voice increasesbrand recall, but only for small, less well-known brands. Finally, increasingshare-of-voice has a significant effect on the purchase intention for smallbrands.

    So, which is more important for the advertiser who wants to make aneffective campaign: media expenditure or message content strategy.^ Formost effects, message content strategy is more important than the abso-lute and relative number of GRPs. The effects of media expenditure aremore limited. However, more media expenditure positively influencescampaign awareness. Media expenditure can also influence brand recalland purchase intention for small brands.

    AcknowledgementsThe author wishes to thank SPOT (the Foundation for Promoting andOptimising Television Advertisements) for providing the data used in thisresearch.

    ReferencesAaker, D.A. & Carman, J.M. (1982) Are you overadvertising? A review of advertising-

    sales studies. Journal of Advertising Research, 22(August), pp. 57-70.Batra, R. & Stayman, D.M. (1990) The role of mood in advertising effectiveness.

    Journal of Consumer Research, YI, pp. 203-214.Batra, R., Lehmann, D.R., Burke, J. & Pae, J. (1995) When does advertising have an

    impact.'' A study of tracking data. Journal of Advertising Research, 35(September),pp. 19-32.

    Belch, G.E., Belch, M.A. & Villarreal, A. (1987) Effects of advertisingcommunications: a review of research, in Sheth, J.N. (ed.) Research in Marketing,Volume 9. Greenwich, CO: Jai Press, pp. 59-117.

    Bemmaor, A.C. & Mouchoux, D. (1991) Measuring the short-term effect of in-storepromotion and retail advertising on brand sales: a factorial experiment. Journal ofMarketing Research, 28, pp. 202-214.

    688

  • WHAT MATfERS MOST IN ADVERTISING CAMPAIGNS?

    Binet, L. & Field, P. (2007). Marketing in the Era of Accountability. Henley-on-Thames,UK: World Advertising Research Center,

    Buzzell, R.D. (1964) Predicting short-term changes in market share as a function ofadvcTUs'mgstrategy. Journal of Marketing Research, 1, pp. 27-31.

    Ehrenberg, A.S.C, (1974) Repetitive advertising and the consumer. Journal ofAdvertising Research, 14(April), pp. 25-34.

    Ehrenberg., A.S.C, Barnard, N. & Scriven, J. (1997) Justifying our advertisingbudgets. Marketing and Research Today, 25(Eebruary), pp. 38^4 .

    Franzen, G. (1999) Brands & Advertising: How Advertising Effectiveness Influences BrandEquity. Henley-on-Thames, UK: Admap Publications.

    Gibson, L.D. (1996) What can one TV exposure o} Journal of Advertising Research,36(March), pp, 9-18.

    Hoyer, W,D. & Brown, S.P (1990) Effects of brand awareness on choice for acommon, repeat-purchase pmauct. Journal of Consumer Research, 17, pp. 141-148.

    Hu, Y, Lodish, L.M. & Krieger, A.M. (2007) An analysis of real world TV advertisingtests: a 15-year upante. Journal of Advertising Research, 47(September), pp, 341-353.

    Jaccard, J,, Turrisi, R, & Wan, C.K, (1990) Interaction effects in multiple regression. SageUniversity Paper Series on Quantitative Applications in the Social Sciences,07-072, Newbury Park, CA: Sage.

    Jones, J.P, (1995a) When Ads Work: New Proof that Advertising Triggers Sales. New York:Lexington Books,

    Jones, J.P. (1995b) Single-source research begins to fulfil its promise. Journal ofAdvertising Research, 3(May), pp. 9-16,

    Laskey, H.A., Fox, R.J, & Crask, M.R. (1995) The relationship between advertisingmessage content strategy and television commercial effectiveness. Journal ofAdvertising Research, 35(March), pp, 31-39.

    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 ofMarketing Research, 32, pp. 125-139,

    MacDonald, E.K. & Sharp, B.M. (2000) Brand awareness effects on consumerdecision making for a common, repeat purchase product: a replication. Journal ofBusiness Research, 48(April), pp. 5-15.

    Maclnnis, D.J,, Rao, A,G, & Weiss, A.M. (2002) Assessing when increased mediaweight of real-world advertisements helps sales. Journal of Marketing Research, 39,pp. 391^07,

    McDonald, C. (1996) Advertising Reach and Frequency: Maximizing Advertising ResultsThrough Effective Frequency. Lincolnwood, IL: NTC Business Books,

    Muehling, D.D, & McCann, M, (1993) Attitude toward the ad: a review. Journal ofCurrent Issues and Resear-ch in Advertising, 15(Fall), pp. 25-58.

    Rossiter, J.R. & Percy, L. (1998) Advertising Communications & Promotion Management(International Edn), Lexington, MA: Lexington Books,

    Rossiter, J.R., Percy, L. & Donovan, R.J. (1991) A better advertising planning grid.Journal of Advertising Research, 31(October), pp. 11-21.

    689

  • INTERNATIONAL JOURNAL OF ADVERTISING, 2009, 28(4)

    Stewart, D.W. & Furse, D.H. (1986) Effective Television Advertising: A Study of 1000Commercials. Lexington, MA: Lexington Books.

    Taylor, R.E. (1999) A six-segment message content strategy wheel. Journal ofAdvertising Research, 39(November), pp. 7-17.

    Tellis, G.J. (1988) Advertising exposure, loyalty, and brand purchase: a two-stagemodel of choice. Journal of Marketing Research, 25, pp. 134-144.

    Tellis, G.J., Chandy, R.K., Maclnnis, D. & Thaivanich, P. (2005) Modeling themicroeffects of television advertising: Which ad works, when, where, for how long,and why.'' Marketing Science, 24, pp. 359-366.

    Trends in Advertising Expenditure (2007) European Advertising & Media Eorecast,21(April, supplement), pp. 9-127.

    Van den Putte, B. (2002) An integrative framework for effective communication, inBartels, G. & Nelissen, W. (eds) Marketing for Sustainability: Towards TransactionalPolicy-Making. Amsterdam: IOS Press, pp. 83-95.

    Van den Putte, B. (2006) A comparative test of the effect of communication strategy,media presence, and previous purchase behaviour in the field of fast movingconsumer goods, in Diehl, S. & Terlutter, R. (eds) International Advertising andCommunication: Current Insights and Empirical Findings. Wiesbaden, Germany:Deutscher Universittsverlag, pp. 89-105.

    Van den Putte, B. & Dhondt, G. (2005) Successful communication strategies: a testof an integrated framework for effective communication. Journal of Applied SocialPsychology, 35, pp. 2399-2420.

    Zhang, Y. & Zinkhan, G.M. (2006) Responses to humorous ads: does audienceinvolvement mdXtex^ Journal of Advertising, 35, pp. 113-127.

    About the authorDr Bas van den Putte is Associate Professor in Commercial Communicationand Health Communication at the Amsterdam School of CommunicationsResearch (ASCoR), University of Amsterdam. His main interest in com-mercial communication is studying what advertising content strategiescan best be used in what situation. His interest in health communicationis in the effect of mass-mediated health campaigns, especially in the fieldof smoking, alcohol and cannabis use. He has published in journals suchas Addictive Behaviors, British Journal of Social Psychology, Journal of AppliedSocial Psychology, Patient Education and Counseling, Psychology of AddictiveBehaviors and Preventive Medicine.

    Address correspondence to: Bas van den Putte, ASCoR, Departmentof Communication, University of Amsterdam, Kloveniersburgwal 48,1012 CX Amsterdam, The Netherlands.

    Email: [email protected]

    690