9
Mining messages: Exploring consumer response to consumer- vs. firm-generated ads EKIN PEHLIVAN*, FUNDA SARICAN and PIERRE BERTHON Bentley University, 175 Forest St., Waltham, MA, USA ABSTRACT Social media provide consumers with a platform for interactivity, and interactivity leads to consumer empowerment by providing the consumer with a platform to make their voice heard. This paper contributes to the marketing literature exploring the voice of the consumer in consumer-generated advertisements (CGAs). The objective of this research is to nd ways to measure consumer response to CGAs. We measure whether they differ from rm-generated ads in the responses they elicit and also observe whether they generate different responses by ad type, or if some categories have similar responses. We review CGAs for Apples MacBook Air lightweight laptop and run a text mining application to understand the common themes and conduct text analysis on the responses to both CGAs and rm-generated ads to answer the question Is the source of the advertisement important?Text analytics also enable us to measure consumersattitude toward products, companies, and ads. We then work toward understanding why and under what circumstances CGAs are effective and how companies may handle or foster different types of CGAs. Copyright © 2011 John Wiley & Sons, Ltd. INTRODUCTION Social media provide consumers with a platform for interac- tivity. Although marketers underestimated the possible con- sequences of interactive marketing on the Internet (Deighton & Kornfeld 2009), they understood that the Internet as well as social media would make the consumer more accessible and bring more power to the marketers and their messages. However, interactivity did more, empowering the consumer by making their voice heard. The latest trends in technology and media have enabled consumers not only to voice their opinions through blogs, podcasts, and various websites but also to engage with their favorite brands by promoting their products and services. One illustration of this empowerment is through consumer-generated advertisements (CGAs) fea- tured on video-sharing websites, such as YouTube. Compa- nies have been confronted with this new consumer behavior recently as technology and media continually present new inventions, products, and services to users. The literature in marketing has recently begun exploring this phenomenon, formulating a denition of the concept and types of CGAs and the responses of companies to them (Berthon et al., 2008). This paper extends this literature by analyzing the extent of consumer power in the marketplace by the use of CGAs. Previous literature looked at the source effects of CGAs using experimental studies (Steyn et al., 2010). We choose a grounded theory approach, taking advantage of the abundance of untapped information on the Web to help in theory building. For our data, we review CGAs for Apples MacBook Air lightweight laptop and run a text mining application to understand the common themes. We then work toward understanding why and under what circumstances CGAs are effective and how companies may handle or foster different types of CGAs. Therefore, the objective of this research is to nd ways to measure consumer response to CGAs. We measure whether they differ from rm-generated ads (FGAs) in the responses they elicit. We also observe whether they generate different responses by ad type (we use four categories of CGAs) or if some categories have similar responses. We conduct text analysis on the responses to both CGAs and FGAs to answer the question Is the source of the advertisement important?Text analytics also enable us to measure consumersattitude toward products, companies, and ads. We conclude by offering managerial strategies based on our ndings. SOCIAL MEDIA AND THE EMPOWERMENT OF THE CONSUMER Traditional marketing efforts are only one method of com- munication from the marketer to the consumer. Advertising was viewed as a way to convey information, invoke emotion, and generate attitude and behavior in the consumer. How- ever, targeting specic consumers was a challenge. Internet advertising, such as banners, pop-ups, or pop-unders, merely emulated this traditional mass media approach. Although the medium was different, the techniques employed in reaching consumers were essentially the same. Then came social media outlets such as Facebook, Twitter, and YouTube, which changed the marketing landscape (Berthon et al., 1996; Watson et al., 2002; Zinkhan & Gelb, 1990). Consumers started voicing their opinions about products or brands through reviews, microblogs, and video blogs, and from these emerged CGAs. This new trend in consumer behav- ior requires a change in the traditional marketing perspectives of companies, since consumers are no longer mere passive receivers of advertisements. Consumers want (and are now able) to get involved in the design and transmission of advertisements. The term vigilante marketing exemplies this trend as unpaid advertising and marketing efforts, including one-to-one, one-to many, and many-to-many commercially *Correspondence to: Ekin Pehlivan, Bentley University, 175 Forest St., Waltham, MA 02452, USA. E-mail: [email protected] Copyright © 2011 John Wiley & Sons, Ltd. Journal of Consumer Behaviour, J. Consumer Behav. 10: 313321 (2011) Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/cb.379

Mining messages: Exploring consumer response to consumer- vs. firm-generated ads

Embed Size (px)

Citation preview

Page 1: Mining messages: Exploring consumer response to consumer- vs. firm-generated ads

Mining messages: Exploring consumer response to consumer- vs.firm-generated ads

EKIN PEHLIVAN*, FUNDA SARICAN and PIERRE BERTHON

Bentley University, 175 Forest St., Waltham, MA, USA

ABSTRACT

Social media provide consumers with a platform for interactivity, and interactivity leads to consumer empowerment by providing theconsumer with a platform to make their voice heard. This paper contributes to the marketing literature exploring the voice of the consumerin consumer-generated advertisements (CGAs). The objective of this research is to find ways to measure consumer response to CGAs. Wemeasure whether they differ from firm-generated ads in the responses they elicit and also observe whether they generate different responsesby ad type, or if some categories have similar responses. We review CGAs for Apple’s MacBook Air lightweight laptop and run a textmining application to understand the common themes and conduct text analysis on the responses to both CGAs and firm-generated adsto answer the question “Is the source of the advertisement important?” Text analytics also enable us to measure consumers’ attitude towardproducts, companies, and ads. We then work toward understanding why and under what circumstances CGAs are effective and howcompanies may handle or foster different types of CGAs.Copyright © 2011 John Wiley & Sons, Ltd.

INTRODUCTION

Social media provide consumers with a platform for interac-tivity. Although marketers underestimated the possible con-sequences of interactive marketing on the Internet (Deighton& Kornfeld 2009), they understood that the Internet as wellas social media would make the consumer more accessibleand bring more power to the marketers and their messages.However, interactivity did more, empowering the consumerby making their voice heard. The latest trends in technologyand media have enabled consumers not only to voice theiropinions through blogs, podcasts, and various websites butalso to engage with their favorite brands by promoting theirproducts and services. One illustration of this empowermentis through consumer-generated advertisements (CGAs) fea-tured on video-sharing websites, such as YouTube. Compa-nies have been confronted with this new consumer behaviorrecently as technology and media continually present newinventions, products, and services to users.

The literature in marketing has recently begun exploringthis phenomenon, formulating a definition of the conceptand types of CGAs and the responses of companies to them(Berthon et al., 2008). This paper extends this literature byanalyzing the extent of consumer power in the marketplaceby the use of CGAs. Previous literature looked at the sourceeffects of CGAs using experimental studies (Steyn et al.,2010). We choose a grounded theory approach, takingadvantage of the abundance of untapped information on theWeb to help in theory building. For our data, we reviewCGAs for Apple’s MacBook Air lightweight laptop and runa text mining application to understand the common themes.We then work toward understanding why and under whatcircumstances CGAs are effective and how companies mayhandle or foster different types of CGAs.

Therefore, the objective of this research is to find ways tomeasure consumer response to CGAs. We measure whetherthey differ from firm-generated ads (FGAs) in the responsesthey elicit. We also observe whether they generate differentresponses by ad type (we use four categories of CGAs) orif some categories have similar responses. We conduct textanalysis on the responses to both CGAs and FGAs to answerthe question “Is the source of the advertisement important?”Text analytics also enable us to measure consumers’ attitudetoward products, companies, and ads. We conclude byoffering managerial strategies based on our findings.

SOCIAL MEDIA AND THE EMPOWERMENT OFTHE CONSUMER

Traditional marketing efforts are only one method of com-munication from the marketer to the consumer. Advertisingwas viewed as a way to convey information, invoke emotion,and generate attitude and behavior in the consumer. How-ever, targeting specific consumers was a challenge. Internetadvertising, such as banners, pop-ups, or pop-unders, merelyemulated this traditional mass media approach. Although themedium was different, the techniques employed in reachingconsumers were essentially the same. Then came socialmedia outlets such as Facebook, Twitter, and YouTube,which changed the marketing landscape (Berthon et al.,1996; Watson et al., 2002; Zinkhan & Gelb, 1990).

Consumers started voicing their opinions about products orbrands through reviews, microblogs, and video blogs, andfrom these emerged CGAs. This new trend in consumer behav-ior requires a change in the traditional marketing perspectivesof companies, since consumers are no longer mere passivereceivers of advertisements. Consumers want (and are nowable) to get involved in the design and transmission ofadvertisements. The term vigilante marketing exemplifies thistrend as “unpaid advertising and marketing efforts, includingone-to-one, one-to many, and many-to-many commercially

*Correspondence to: Ekin Pehlivan, Bentley University, 175 Forest St.,Waltham, MA 02452, USA.E-mail: [email protected]

Copyright © 2011 John Wiley & Sons, Ltd.

Journal of Consumer Behaviour, J. Consumer Behav. 10: 313–321 (2011)Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/cb.379

Page 2: Mining messages: Exploring consumer response to consumer- vs. firm-generated ads

oriented communications, undertaken by brand loyalists onbehalf of the brand” (Muñiz and Schau 2007, 187). Of course,this is a double-edged sword: consumers are free to generateads that portray both positive and negative views of productsand firms (Berthon et al., 2007; Pitt et al., 2002).

Berthon et al. (2008) define CGAs as “any publicly dis-seminated, consumer generated advertising message whosesubject is a collectively recognized brand/product.” Theygo on to identify four types of CGA based on the relationshipof the message to the FGA (congruous/incongruous) and thetone of the CGA (positive/negative). When a CGA is inaccordance with the official message and a positive tonetoward the brand/product, it is termed a concordant ad. Onthe other hand, a subversive ad would be in general agree-ment with the official message but with an underlyingnegative attitude. Incongruous ads do not convey the officialmessage of the brand but nevertheless have positive attitudes.Finally, CGAs that are off the message and clearly have anegative tone are labelled contrarian ads.

A common expectation has been that consumers wouldview CGAs as being honest and sincere and hence moreeffective than FGAs. Steyn et al. (2010) found that the pop-ularity of CGAs was the only significant variable to createpositive feelings. This paper looks at field data to see whetherCGAs have a different appeal than FGAs and whether theireffectiveness differs within the four categories proposed byBerthon et al. (2008).

METHODOLOGY AND DATA

Data and codingThe data collected for this study consists of comments postedin response to the official ads and the CGAs for the AppleMacBook Air. We selected this product since the brand hasa very active community in terms of creating videos andposting comments. Further, the comments often containinformation pertaining not only to the product and brandbut also to the creative quality of the videos. Therefore, thesedata enable a multidimensional analysis of what the consu-mers think about MacBook Air, Apple, its competitors, andthe specific ad.

We looked at 75 ads, 49 of which matched our criteria ofbeing less than 2 minutes long and relevant to the product.Two independent researchers coded the ads based on thedefinitions of each type provided by Berthon et al. (2008).The process included re-viewing of the 49 chosen videos toanswer three questions pertaining to the creator (FGA vs.CGA), message (congruous vs. incongruous), and attitude(positive/negative) as binary variables. Combinations of eachbinary variable create the category; “if” statements in Excelwere formulated to label and categorize each observation.Table 1 shows the number of videos by category (FGA andthe four types of CGA) along with the total number ofcomments that each ad elicited. Of the official FGAs, therewere 4 videos with 887 comments. Of the CGAs, 12 wereincongruous with 109 comments, 14 concordant with 298comments, 3 contrarian with 3716 comments, and 20 subver-sive with 8570 comments.

The 13,580 comments were analyzed using text mining.Because manual analysis of such large amounts of unstruc-tured data is impractical, text mining can help extract themesand patterns; mining can reveal associations and relation-ships among terms as well as identify trends (Lee et al.,2010; Davi et al., 2005). We employed SAS Text Miningand NVivo in combination to help understand how theresponses toward advertisements differed from each other.We used the packages to mine for main themes in the dataand to discover relationships and differences between CGAsand FGAs. Using SAS Text Miner and NVivo, we ran ouranalysis on the comments for each type of ad. We excludedirrelevant terms such as user IDs, dates, numbers, addresses,prepositions, abbreviations, and noninformative parts ofspeech. In addition, we united synonyms of the words intoone word.

The results are summarized in the next section, followingthat we discuss the results to reach a comprehensive under-standing of the comments. Then we follow up with theoreti-cal and managerial implications and suggestions for futureresearch.

RESULTS

Our analyses consist of word frequency queries and conceptlink searches. Word frequency queries provide a list of themost frequently appearing words in the selected documents.Word frequency queries are used to identify major themesin the data. The “concept link search” feature in SAS TextMiner generates a link from a first word to a second wordwhen the second word occurs with the first at least 5 per centof the time.

Word frequency query resultsThe figures in Table 2 show the most frequent words alongwith length (the number of letters in the word), count (thenumber of times that the word has occurred), and percentage(the frequency of the word relative to the total word counted)(see NVivo, 2011, for software details). The “Similar words”column shows the words that are treated as synonymous (i.e.,the word stem is the same, for example, mac and macs). Wefirst looked at the frequency of different words in commentson FGAs and CGAs. In both cases, “macs,” “macbooks,”“having,” “like,” “air,” “apple,” “getting,” “just,” “pc,”“computer” are the most frequently used words (see Table 2).

We extended this analysis by splitting the CGAs into theirfour subtypes (contrarian, incongruous, subversive, and con-cordant). “Laugh out loud” (abr. lol), “air,” “have,” “laptop,”

Table 1. Summary of data

Category Number of ads Number of Comments

FGAs 4 887CGA-Concordant 14 298CGA-Subversive 20 8570CGA-Incongruous 12 109CGA-Contrarian 3 3716

314 E. Pehlivan et al.

Copyright © 2011 John Wiley & Sons, Ltd. J. Consumer Behav. 10: 313–321 (2011)

DOI: 10.1002/cb

Page 3: Mining messages: Exploring consumer response to consumer- vs. firm-generated ads

“like,” “nice,” and “macbook” are the most common wordsin the concordant category. Similarly, “lol,” “mac,” “funny,”“great,” “really,” “ad,” “macbook,” and “air” are the mostfrequent words in the incongruous category. “Macs,”“having,” “lols,” “like,” “macbooks,” and “air” are the mostfrequent words in the subversive category. In contrast,contrarian ads received comments involving “thin,” “how,”“know,” “lol,” “envelope,” and “ad” most frequently(Table 3).

In general, CGAs elicit positive attitudes as viewers usewords such as “lol,” “funny,” and “like” in their comments.Furthermore, while most of the comments talk about the

Table

2.FGA

andCGA

wordfrequencies

FGA

CGA

Word

Length

Count

Weightedpercentage

(%)

Sim

ilarwords

Word

Length

Count

Weightedpercentage

(%)

Sim

ilarwords

Macs

4164

2.03

Mac,m

acs

Macs

4325

1.42

Mac,macs

Macbook

7128

1.58

Macbook,m

acbooks

Lol

3317

1.38

Lol,lols

Having

6110

1.36

Have,having

Macbooks

8284

1.24

Macbook,m

acbooks

Apple

591

1.13

Apple,apples

Having

6265

1.16

Have,haves,having

Like

489

1.10

Like,lik

ed,likes

Airs

4250

1.09

Air,airs

Gettin

g7

881.09

Get,g

ets,getting

Liking

6204

0.89

Like,lik

ed,likes,lik

ing

Air

381

1.00

Air

Just

4183

0.80

Just

Pc

279

0.98

Pc

Whats

5162

0.71

What,whats

Has

376

0.94

Has

Gettin

g7

157

0.69

Get,’get,gets,g

ettin

gLaptop

671

0.88

Laptop,laptops

Laptop

6147

0.64

Laptop,

laptopes,laptops

Just

463

0.78

Just

Funny

5144

0.63

Funni,funny

Using

560

0.74

Use,used,useful,uses,using

Com

putin

g9

141

0.62

Com

puter,computers,com

puting

What

460

0.74

What,whats

Has

3141

0.62

Has

Song

459

0.73

Song,

songs

Using

5135

0.59

Use,u

sed,

useful,u

ses,using

Com

puter

856

0.69

Com

puter,computers,com

puting

Pc

2134

0.59

Pc

Drives

656

0.69

Drive,d

rives

All

3130

0.57

All,

all’

Better

653

0.66

Better

Apples

6130

0.57

Apple,apples

Table 3. Top 10 most frequent words in concordant, contrarian,incongruous, and subversive categories

Word Length CountWeighted

percentage (%) Similar words

ConcordantLol 3 59 2.78 LolAir 3 33 1.55 AirHave 4 32 1.51 Have, havingMacbooks 8 30 1.41 Macbook, macbooksLaptop 6 25 1.18 Laptop, laptopes,

laptopsLike 4 24 1.13 Like, likingNice 4 24 1.13 Nice, nicelyThinkpad 8 20 0.94 Thinkpad, thinkpadsHaha 4 17 0.80 HahaJust 4 17 0.80 JustContrarianThin 4 4 3.51 ThinHow 3 3 2.63 HowKnow 4 3 2.63 KnowLol 3 3 2.63 LolXd 2 3 2.63 XdEnvelope 8 2 1.75 EnvelopeFrom 4 2 1.75 FromFuck 4 2 1.75 Fuck, fuckingHahaha 6 2 1.75 HahahaMean 4 2 1.75 MeanIncongruousLol 3 11 2.28 LolMac 3 8 1.66 MacMacbook 7 8 1.66 MacbookAir 3 7 1.45 Air, airsFunny 5 7 1.45 FunnyGreat 5 7 1.45 GreatReally 6 7 1.45 ReallyAd 2 6 1.24 AdComputers 9 6 1.24 Computer, computersJust 4 6 1.24 JustSubversiveMacs 4 253 1.49 Mac, macsMacbooks 8 220 1.30 Macbook, macbooksHaving 6 211 1.24 Have, haves, havingAir 3 185 1.09 Air, airsLols 4 159 0.94 Lol, lolsJust 4 147 0.87 JustLike 4 146 0.86 Like, liked, likes,

likingGetting 7 133 0.78 Get, ’get, gets,

gettingWhats 5 121 0.71 What, whatsLaptop 6 114 0.67 Laptop, laptopes,

laptops

Response to consumer- vs. firm-generated ads 315

Copyright © 2011 John Wiley & Sons, Ltd. J. Consumer Behav. 10: 313–321 (2011)

DOI: 10.1002/cb

Page 4: Mining messages: Exploring consumer response to consumer- vs. firm-generated ads

product, the ad itself is the second most important theme inthe comments.

Concept link search resultsAfter analyzing word frequencies, we run concept links toidentify the relationships between terms. We ran the “conceptlink” search for words that identified the themes within thecomments. In each figure, the central theme is positioned atthe center; the peripheral links are the concepts that occur withthe central theme commonly. For instance, Figure 1a demon-strates a SAS Text Miner “concept link” search for the word“air” that is a central theme in FGAs (see SAS, 2011, for soft-ware details). It shows that “air” is linked to words such as“+drive,” “fan,” “+inch,” “+have,” “macbook,” “+apple,”“+make,” and “+see.”

Similarly, Figures 1b, 1c, and 1d illustrate the relation-ships of themes dominating within the comments for theFGAs, like “macbook,” “song,” and “apple,” respectively.

On the one hand, in Figure 1b, concept links for“macbook” indicate that comments related to the productare mostly descriptive.

On the other hand, Figure 1d lists attitudinal connectionsto the brand. We can say that viewers are polarized sincethe brand name has a relationship with both of the words“love” and “hate.” Further, comments include many Apple

products. Viewers have knowledge about the variety of itsproducts.

ConcordantConcordant ads share the official message and the positiveattitude that FGAs also have. Most comments in these areabout the product and its price.

As seen in Figure 2, the viewers emphasize the size andthe design of the product. There is also concern about priceas “air” is associated with “money,” “+buy,” and “spend.”

ContrarianContrarian ads deviate from the official message and have anegative attitude. These types of ads were least frequentand yet generated the most number of comments. Thecomments were distributed on a wide range. Figure 3a showsthat the viewers of contrarian ads are discussing the productfeatures and the price as well as the ad.

As Figures 3b and 3c illustrate, the words “+video” and“+song” generate appreciation and inquiry. Viewers requestto know what the name of the song and the artist are.

IncongruousWhen an ad features a different message than the FGA andhas a positive attitude toward the product, it is categorized

a: Air Concepts Links

c: Song Concepts Links d: Apple Concepts Links

b: Macbook Concepts Links

Figure 1. FGA concept links.

316 E. Pehlivan et al.

Copyright © 2011 John Wiley & Sons, Ltd. J. Consumer Behav. 10: 313–321 (2011)

DOI: 10.1002/cb

Page 5: Mining messages: Exploring consumer response to consumer- vs. firm-generated ads

Figure 2. Concordant concept links.

a: Macbook Concept Links b: Video Concept Links

c: Song Concept Links

Figure 3. Contrarian concept links.

Response to consumer- vs. firm-generated ads 317

Copyright © 2011 John Wiley & Sons, Ltd. J. Consumer Behav. 10: 313–321 (2011)

DOI: 10.1002/cb

Page 6: Mining messages: Exploring consumer response to consumer- vs. firm-generated ads

as incongruous. In the MacBook Air example, incongruousads featured themes about the ad. More specifically, the word“ad” had links with words “great” and “job,” suggesting thatthe ad is appreciated among the viewers (see Figure 4.)

SubversiveSubversive ads have the official message but a negativeattitude toward the product/brand advertised. The commentsfor this type of ads were focused on the ad like the otherCGAs and the product itself. In Figure 5, the product“mac” had association with the words “+pc,” “os,” and“windows.” Viewers of these ads compared the software,and there is a very clear polarization in the comments, ascan be seen in the examples below:

“I am definitely a mac person by cool ad.”“Windows is no more or less secure than a Mac.”“Well, let me say that I have been a PC guy for 20 years.”

DISCUSSION

Our analysis identified patterns in ad responses in terms offrequency and relationships. Both the categories of FGAsand CGAs and the types of CGAs among themselves sharesome common themes of emphasis and differ from eachother at other points of focus. The comments in their entiretyare a testament to how the contemporary consumer becomes

Figure 4. Incongruous concept links.

Figure 5. Subversive concept links.

318 E. Pehlivan et al.

Copyright © 2011 John Wiley & Sons, Ltd. J. Consumer Behav. 10: 313–321 (2011)

DOI: 10.1002/cb

Page 7: Mining messages: Exploring consumer response to consumer- vs. firm-generated ads

vocal in cyberspace and how they are empowered throughsocial media.

Figure 6 illustrates main themes with example words.Three main themes emerged for FGA comments: product/brand, competitor’s product, and the ad. Similarly for CGAs,the main themes were the product/brand, the competition,and the ad itself. Within the comments focusing on the adin the CGA category, there were attitude and descriptivepatterns.

As the work sets in Figure 6 reveal, the main themes aresimilar whether the ads are firm generated or consumergenerated. What stands out, however, is the differencebetween the content discussed within the theme. Theviewers express appreciation for the FGA and emphasizethe song; however, they find the CGAs funny and talkmore about the visuals such as the envelope within the adtheme. The product/brand is described in further detail inthe FGA comments in comparison to the comments leftfor the CGAs.

In terms of being a catalyst for discussion, contrarian adsstood out with 3716 comments for just three ads. By defini-tion, contrarian ads deviate from FGAs in terms of both themessage and the attitude. One may conclude that this devia-tion from the original ad creates a greater space for discus-sion. The ad that got most of the comments was a comedictake on eating disorders of a fat laptop (PC) as she sees a thinone (MacBook Air). Interestingly, most comments focusedon the ad rather than the product itself. For instance, onecomment explained the plot to the rest of the viewers whoseemed to be lost:

“ok. you see hes watching thin notebook? hes not thin. soyou see the food next to it all the time? it has eatingproblems. it watched the fat girl and tried to work out.Then it became anorexic and throwing up to be skinnylike the MacBook. GET IT YET?”

Other comments praised the ad and the creator:

“Very, very good :)”“My favorite part is when it’s doing sit-ups. (Open-and-closes?)”“That was so cool!!!! I’m laughing my head off!!!!”

Some had information for those who wished to buyMacBook Air:

“thalliumproductions is giving away an macbook forfree. all you have to do is be subscribed to him, theresa video with all the info in my channel. so subscribeto him”

The word frequency query revealed several interestingpoints; the similarities and differences between the com-ments left for FGAs and CGAs were rather curious. First,references to the product itself, such as Mac, MacBook,Apple, and Air, as well as references to the category ofPCs, computers, or laptops are among the common wordsfor both categories. Secondly, allusion to the song usedin the original ad is mostly seen in the FGAs. Audienceasks for the name of the song and the singer, and othersrespond with the information. It is worth noting that thesame song is used in many of the CGAs; however, thenewness factor probably diminished after the original adsaired there for references in the CGAs, although notnonexistent, became less. Also, in the comments for FGAs,we encounter comparatives such as “better” or “more” andaction verbs such as “buy,” which is not as common in thecomments left for CGAs. Finally, CGA comments featureabbreviations such as “LOL,” smileys like “XD,” anddescriptives for the ad such as “funny,” “great,” and “nicejob” to indicate the general attitude toward the CGAs.From these we deduce that humor is more commonly usedin CGAs and that the audience is able to appreciate it.Furthermore, the song choice on the part of Apple provesto be successful in creating a light mood to get theconsumer into a buyer’s mindset (Van Raaij 1993). Onemight even argue that it is perhaps a bit too successfulinasmuch as the song overshadows the product at timesin the discussion.

The last stage of our analysis focused on the relation-ship between the most commonly occurring words andthemes. For this, we used the concept link search tooland found that for FGAs, the comments are productdescriptive. The concept links on Apple especially revealthat the audience has knowledge of Apple products, notlimited to the product featured in the ad. Moreover, thelinks between the brand “Apple” and both “like” and

Ads

Themes

Patterns

Examples

Figure 6. Themes and patterns.

Response to consumer- vs. firm-generated ads 319

Copyright © 2011 John Wiley & Sons, Ltd. J. Consumer Behav. 10: 313–321 (2011)

DOI: 10.1002/cb

Page 8: Mining messages: Exploring consumer response to consumer- vs. firm-generated ads

“hate” are strong, suggesting that there may be polariza-tion in the discussion. Several such comments carry thediscussion as follows:

“this made me fall in love with apple products..”“I like Mac computers. I HATE mac users :D I also lovemy PC laptop. macbook user trolls and L2play mine-sweeper.”“the only reason mac people hate pc people is becausethey spent a fortune to buy the crappy OSX that can donothing serious, not even play games, while pc users get5 times more things with 1/3 the price. . .”

Another result from the concept link search revealed thatcomments pertaining to the CGAs of all types focus ondescribing the ad. The description is provided by wordassociations such as “video” or “ad” and “funny.” Positiveattitude toward the incongruous ads, for instance, are evidentin the links between “ad” and “great,” “nice,” and “job.”Two categories of CGAs, the concordant and the subversiveones, also feature links that focus on the product. Forinstance, there exist concept links between “apple” or “air”and “thinkpad” pointing to a comparison. Also, there arereferences to the price of the product and features such asUSB “drives” (or lack thereof).

The examples and results mentioned above all supportthe premise that the social media provide the venue forcustomers to let their opinions be known about anythingand everything. The comments’ themes, as suggestedearlier, range from the product/brand or the competitionto the ad itself or the creator of the ad. There is not muchinhibition when it comes to what consumers say or evenhow they choose to say it, whether it be through videosas in the example of CGAs or text as in the example ofcomments.

CONCLUSION

“Three of the world’s most popular brands online aresocial-media related (Facebook, YouTube and Wikipedia)and the world now spends over 110 billion minutes onsocial networks and blog sites. This equates to 22 percentof all time online or one in every four and half minutes.”(Neilsen Reports, 2010)

In consideration of the significant impact that socialmedia have had on how we relate to each other, it is impor-tant once more to realize its impact on marketing and theconsumer. We find that consumers are empowered throughthe freedom that social media provide when it comes tomaking the consumer’s voice heard. In this paper, weanalyzed a set of comments left for FGAs and CGAs forthe MacBook Air. Using a grounded theory approach, weexplored source effects when it came to the responsestoward ads, and on which topics consumers were vocal.We found that the comments differ for the FGAs andCGAs; there was no indication of a source effect but ratherreferences to the dominant features such as the song for the

FGAs and the humor for the CGAs. We also found patternswithin the comments about which types of ads and topicselicited discussion from consumers. Ads that deviatedfrom the FGAs, both in the message and the attitude, werethe ones that generated the most comments. We believethis is because the deviation provides originality and morepoints of discussion. When it comes to the common themesand patterns in the comments, we see that the product,MacBook Air; its alternatives, Thinkpad, laptops, PCs;and their features are one identifiable group. The otherpatterns that emerge in the CGAs are mostly about how“funny” or “good” the ads are.

Limitations and directions for future researchDuring the analyses, we encountered several limitationsto our chosen methodology. The first one was concerningthe coding of CGAs; however, the existence of two inde-pendent coders and the binary system focusing on thefeature of the ads rather than the ads themselves helpedalleviate this limitation. Another one was choosing theappropriate tools of text mining among many options.We decided to use SAS Enterprise Text Miner and NVivoafter careful consideration and a survey of the literature;both programs provide us with two important analyses:word frequency queries and concept link searches. Thethird important limitation of this study was in part aboutmaking sense of the results because we chose a groundedtheory approach. Our starting point was not a theory butthe observations from the data we gathered; hence, theresults were the fruit of discovering the themes and patterns,not identifying what was suggested by previous researchers.Since all cases would be unique in terms of the results,starting from the data is more sensible than starting withpreconceptions.

Future research should focus on generalizable patternsof response to ads from different sources. This way, asystematic strategy can be formulated. Our results sug-gest that there are three important topics that receive aresponse: the product, the competitors, and the ad. Withenough data, it might be possible to create a model toexplain the role that each topic plays and test the modelto see which ones are more influential.

Scholarly and managerial contributionsThis paper contributes to the literature by analyzing theextent of consumer power in the marketplace by theuse of CGAs as well as source effects as manifested inthe comments, using a grounded theory approach. To ourknowledge, the literature does not have examples of thissort. Our analysis identified patterns in ad responses andenabled us to provide suggestions and recommendationson how to use this information.

The managerial implications of these findings rest in thesubtle differences between the responses to FGAs andCGAs. With the growth of social media, companies mayinflict unnecessary harm to their businesses by ignoringCGAs and the impact of communicative technology on their

320 E. Pehlivan et al.

Copyright © 2011 John Wiley & Sons, Ltd. J. Consumer Behav. 10: 313–321 (2011)

DOI: 10.1002/cb

Page 9: Mining messages: Exploring consumer response to consumer- vs. firm-generated ads

products. Companies must recognize this new trend inconsumer behavior and should capitalize on consumer effortsto promote their brands by adjusting their marketing strate-gies to harness this trend.

REFERENCES

Berthon PR, Pitt LF, McCarthy I, Kates SM. 2007. When customersget clever: managerial approaches to dealing with creativeconsumers. Business Horizons 50(1): 39–47.

Berthon PR, Pitt LF, Campbell C. 2008. Ad lib: When customerscreate the ad. California Management Review 50(4): 6–30.

Berthon PR. Pitt LF, Watson RT. 1996. Marketing communicationand the world wide web. Business Horizons 39(5): 24–32.

Davi A, Haughton D, Nasr N, Shah G, Skaletsky M, Spack R. 2005.A review of two text-mining packages: SAS Text Mining andWordStat. The American Statistician 59(1): 89–103.

Deighton J, Kornfeld L. 2009. Interactivity’s unanticipated conse-quences for marketers and marketing. Journal of InteractiveMarketing 23(1): 4–10.

Lee S, Song J, Kim Y. 2010. An empirical comparison of four textmining methods. The Journal of Computer Information Systems51(1): 1–10.

Muñiz AM, Schau HP. 2007. Vigilante marketing and consumer-created communications. Journal of Advertising 36(3): 187–202.

Neilsen Reports, 2010. Available at http://blog.nielsen.com/nielsenwire/online_mobile/social-media-accounts-for-22-percent-of-time-online/ [accessed on 07 June 2011].

NVivo. 2011. Using the Software. Available at http://help-nv8-en.qsrinternational.com/Word_Frequency_Queries.htm [accessed on07 January 2011].

Pitt LF, Berthon PR,Watson RT, Zinkhan GM. 2002. The internet andthe birth of real consumer power. Business Horizons 45(4): 7–14.

SAS. 2011. SAS Text Miner: Use Text Mining to a Analyze Unstruc-tured Data Sources. Available at http://www.sas.com/technologies/analytics/datamining/textminer/ [accessed on 05 January 2011].

Steyn P, Wallström Å, Pitt LF. 2010. Consumer-generated contentand source effects in financial services advertising: an experi-mental study. Journal of Financial Services Marketing 15(1):49–61.

Van Raaij WF. 1993. Postmodern consumption. Journal ofEconomic Psychology 14(3): 541–563.

Watson RT, Pit LF, Berthon PR, Zinkhan GM. 2002. U-Commerce:expanding the universe of marketing. Journal of the Academy ofMarketing Science 30(4): 333–347.

Zinkhan GM, Gelb BD. 1990. Repetition, social settings, perceivedhumor, and wearout. in Goldberg ME, Gorn G, Pollay RW (eds).Advances in Consumer Research 17(1). Association forConsumer Research: Provo, UT; 438–441.

Response to consumer- vs. firm-generated ads 321

Copyright © 2011 John Wiley & Sons, Ltd. J. Consumer Behav. 10: 313–321 (2011)

DOI: 10.1002/cb