The Role of Advertising in Word of iVIouth

ED KELLER The Keller Fay Group ekeller@kellerfay.com BRAD FAY The Keller Fay Group bfay@kellerfay.com

Word of mouth (WOM) is now widely believed to help drive consumer decision making. What lias not been as widely discussed is the relationship between advertising and WOM. New empirical data collected by the authors over a three-year period show that 20 percent of WOM discussions refer to paid advertising In media. Further, the research finds that conversations that are "advertising influenced" in this way are significantly more likely to involve recommendations to buy or try a brand when compared with other WOM discussions about brands. These findings hold across a wide range of product categories.

INTRODUCTION

One of the seminal studies tn mass communications was conducted in 1944 by Columbia University Professor Paul Lazarsfeld. "The Decatur Study," as it came to be known, was conducted among women and sought to understand the role of media and of individuals who influenced recent consumer decisions. The study focused on four categories: "daily household marketing," fashion and beauty, movies, and politics. It found that mass communications campaigns often fail to meet expectations because of "selectivity" in exposure and perception by recipients for whom messages are not of immediate interest or relevance, or because people hold preexisting opinions that may be dissonant with the message. The results of the Decatur Study were later published by Lazarsfeld and Elihu Katz in Personal Influence, in which they describe and validate what they called the "two-stepflow"communications process in which the power of mass media is significantly enhanced by delivering messages to opinion leaders who "mediate" the information and disseminate it to the broader populace (Katz and Lazarsfeld, 1955). Katz and Lazarsfeld concluded that mass commurucations are more apt to reinforce supporters or current customers, rather than convert new people. It was in this context that they found opinion leaders were more inclined to absorb mass 1 5 4 JOUfliiBL OF HDUERTISinG RESEHRCH June 2 0 0 g

media communications and then disseminate them to the general public. Despite these findings, most advertising during the intervening decades has continued to focus on creating awareness, preference, and purchase intent among mass audiences, and on attracting and then converting prospects. Today, this model is coming under increasing scrutiny as marketers question the efficacy of traditional advertising approaches. A number of books have provocatively announced the "end of advertising," due especially to declining trust, the profiliferation of media choice, and the rise of the internet (Godin, 2005; Jaffe, 2005; Ries and Ries, 2002; Semovitz, 2006; Zyman and Brott, 2002). This discussion is further driven by research that illustrates the growing impact of word of mouth (V\/OM) as the primary driver of consumer purchase decisions (Advertising Age, 2008). New research by our firm shows, however, that this is a false dichotomy and that there is an important interaction effect between advertising and WOM, much as was posited by Katz and Lazarsfeld and subsequently by others such as Bayus (1985), In this article, we provide new empirical evidence about the importance of mass communications to WOM. The two, we believe, are not in conflict with each other, but rather work quite effectively when they work together.
DOI: 10.2501/S0021849909090205

ROLE OF ADVERTISING IN WOM

EMPIRICAL GENERALIZATIONS
Approximately 20 percent of WOM about brands refers to paid advertising in media. The ievel and effectiveness of WOM is substantiaiiy increased when stimulated, encouraged, and/or supported by advertising, increasing the probability by about 20 percent that a consumer will make a strong recommendation to buy or try a product.

Respondents are classified based on whether or not they are opinion leadersin our terminology, "Conversation Catalysts^""based on the size of their social network and their self-reported recommending behavior in a variety of product categories. This information all leads to a deeper understanding of the interaction between traditional paid-media advertising and WOM in a way that factors in both offline as well as online conversations, positive as well as negative WOM, and "senders" as well as "receivers" of WOM advice and recommendations.
THE TWO-STEP FLOW THEORY REVISITED

For the purpose of this article, we begin with two empirically derived generalizations; EGl. Approximately 20 percent of WOM refers to paid advertising in media. EG2. These discussions are more likely to involve a recommendation to buy than other WOM about a brand.

Our first generalization regarding the role of advertising in stimulating WOM is important because Katz and Lazarsfeld first postulated the two-step flow at the early stages of the mass media era. More than 50 years later, we see the extent to which that generalization still holds as we enter what some observers believe is the beginning of the post mass media era. Our second generalization is important because it suggests that the rising importance of WOM does not come at the expense of advertising and other mass communications, but rather WOM provides a crucial mechanism through which advertising and marketing works. The implications for mass communication and advertising are discussed at the end of this article, following the presentation of empirical evidence for our two generalizations.
PRIMARY DATA SOURCE

tive sample of 700 Americans ages 13 to 69 about the "conversations" they participated in the day before the interview. The surveys are administered oniine. The participants are presented initially with a two-page diary they use to keep track of conversations in 15 product categories for a single day. The following day they complete a 20-minute questionnaire in which they list the brands that came up in conversation and then answer detailed questions about their conversation regarding each brand. In this article, our findings cover the 12 months ending September 2008, during which time 36,402 respondents were interviewed and data for 294,887 conversational mentions of brands were collected. TalkTrack also collects several types of information that are unique and particularly important for the purposes of this article: Respondents who participated in a conversation about a brand are asked whether they or any other person in the conversation referred to an information source about that brandsuch as advertising, coupons, website, etc, and the type of information source (e.g., TV commercial, magazine article, website, etc.) is recorded. Respondents who were on the "receiving end" of advice or opinions about a brand are asked whether they received an actual recommendation about the brand, and also the likelihood that they will purchase the brand based on the conversation.

To empirically confirm the Katz and Lazarsfeld theory, we would expect to find evidence for the following: a substantial proportion of conversations is influenced by advertising; a process that involves "opinion leaders" as crucial disseminators of WOM about brands, particularly advertisinginfluenced WOM. For the purpose of this discussion, we are confining our definition of "advertising" to paid-media advertising. While TalkTrack also collects information related to public relations, promotion, point of sale, direct marketing, and other forms of marketing and advertising, we will leave aside those marketing channels for future analysis. One of the most important discoveries in our research is that 22 percent of all WOM about brands involve some participant in the conversation referring to something that they saw or heard in paidmedia advertising. In fact, we believe this is a conservative estimate of advertising's role in WOM because it only counts conversations wherein advertising is

The primary source of data for this article is an ongoing series of studies from the Keller Fay Group known as TalkTrack. livery week since June 2006, TalkTrack interviews a fresh, nationally representa-

June 2 0 0 9 JOOHfldL DF HDUEI1TISII0 HESEHHCH 1 5 5

ROLE OF ADVERTiSiNG iN WOM

specifically mentioned by one or more participants. It does not include occasions when an advertisement indirectly (or unconsciously) motivated (or provided content to) a conversation about a brand. Yet even with this conservative estimate, when we apply 22 percent to our estimate of daily WOM impressions, we project 716 million daily conversations about brands that are "advertisinginfluenced." This does not suggest that advertising was the sole cause for the conversation, but orUy that advertising was part of the discussion. It might have been the cause, or it might have been mentioned in response to a discussion stimulated by something else. TalkTrack provides coverage in 15 distinct product categories, allowing for examining the generalizability across industries (see Table 1). We found that more than a fifth of aU WOM is about advertising in 10 of the 15 survey categories, and that for all industries at least 15 percent of WOM is influenced by paidmedia advertising, leading us to conclude this is a law-like pattern. The most influenced industries are entertainment/movies, technology, and beauty. Interestingly, two of these were among the categories studied in the Katz and Lazarsfeld Decatur study. We also found that online WOM communications about brands were more likely to contain references to advertising than offline conversations; 32 percent of online WOM contains advertising references, compared to 21 percent of offline WOM. We hypothesize that because online communications occur in an ad ver tising-rich environment that lends itself to the forwarding of web links, online WOM is more likely to be advertising influenced than offline WOMa difference that may present digital advertisers with a very attractive opportunity to extend the reach of their messages.

TABLE 1 Brand Conversations Influenced by Advertising

Percent of WOM Influenced by Advertising Industry Entertainment/movies Technology Persona! care/beauty Automotive Telecommunications Shopping/retail Household products All category average The home Travel services Food/din ing Children Beverages Sports/recreation/hobbies Financial Health/healthcare 27% 26% 25% 24% 24% 23% 22% 22% 22% 21% 20% 19% 18% 17%
16%

and often at an earlier stage (either after a new-product release or following news/ information about an existing product or marketing activity). We find that efficiency gains from targeting influencers can double the impact of WOM marketing at the low end, and at the high end it can multiply efficiencies by as much as fve times.

ADVERTISING'S ROLE IN THE QUALITY OF WOM

What about the qualit]/ of talk about brands, and the effectiveness of that talk? Many researchers have found that consumers are reluctant to admit that advertising has an impact on their purchasing decisions. Much of the enthusiasm for WOM as a discipline is based on the assumption that "authentic" WOM, which stems from positive brand experiences, is more effective than commercially paid messages. TalkTrack does confirm the power of WOM in driving behavior (see Figure 1). Using a 0 to 10 scale where "0" means

15%

Base: Brand convernalions influenced by advertising, 32,496 Source: Keller <iy Group's TalkTracK*, October 2007 through Sqitembcr 2008

49%

49%

WOM is influenced qualitatively (and most probably quantitatively) by advertising. To fully validate the Katz and Lazarsfeld model, we would also like to test whether a segment reasonably labeled "opinion leaders" is disproportionately involved in the dissemination of advertisinginfluenced WOM. We have undertaken such research, but for reasons of space we carmot include the results here. We have published elsewhere on influencers and WOM (Keller and Fay, 2007). Among the key findings are that influencers engage in significantly more WOM than other consumers across all categories of brands.

Likely to Purchase

Likeiy to Pass Along

Likely to Seek Out Information

Base: Brand Conversations, n = 95.505 Source: Keller Fay Group's TalkTrack. October 2007 through September 2008

Figure 1 Intended Action of Those Receiving WOM Advice/Opinions (Percent Responding 9 or 10 on a Scale of 0-10, Where 10 Means "Extremely Likely")

1 5 6 JDUHOeL or fldOERTISIHG RESEHHCH June 2 0 0 9

ROLE OF ADVERTISING IN WOM

"very unlikely" and "10" means "extremely likely," respondents on the receiving end of conversations about brands say they are highly likely to take action based on the recommendations. As many as 49 percent who answered with a "9" or "10" reported that they are likely to purchase each of the brands they talked ,ibout in their prior-day conversations. Simiiarly, 49 percent said they are likely to pass the advice along to somebody else, and 35 percent said they will seek out additional information about the brands. Although these findings are based on

far more apt to have no recommendation at all (31 percent) compared to advertisinginfluenced WOM (18 percent). Our initial hypothesis is that good advertising reminds people of why they like a particular brand and makes it easier to recommend the brand with a specific rationale attached to that recommendation. In the words of Katz and Lazarsfeld, advertising plays a "reinforcing" function that makes WOM stronger.

We have tested this generalization across the 15 categories covered by TalkTrack and find a positive and statistically significant relationship between WOM that is advertising influenced and strong recommendations being made for 12 of our 15 categories (see Table 2). The difference between advertising and non-advertising-related WOM and recommendations is strongest for entertainment/ movies, financial ser\nces, the home, sports/

IMDLC ^

self-reported intentions, they have been confirmed by evaluating data in TalkTrack'^ against third-party behavioral data
in a recent analysis by Graham and Havlena (2U7). Graham and Havlena found that in the categories studied, there was a clear association between rising advertiso

Prevalence of Strong "Buy/Try" Recommendation within a Brand Conversation, Depending on Whether Conversation Contained a Reference to Advertising*
Percent of WOM Receiving Recommendation to "Buy or Try" -j., 4,. . ^ ., , . * . _
AdverttslngNon-Ad vertising-

ing levels leading to higher WOM levels and ultimately to rising consumer behavior in the form of web searches and site . .,
visits.

Entertainment/movies

47% 35% 47% 34% 60%_ 41% 40% 34% 53%_ 33% 57% ^8% 37% 49% 54% . , *

36% 25% 37% 24% 51% 34% 34% 28% 47% 28% 52% 54% 34% 46% 54% '

+11**
...::

But the question remains: how does .ldvertising-influenced WOM compare in efficacy to other forms of WOM? When advertising influences WOM, do we see different levels of efficacy? By one set of measures, we see that advertisinginfluenced WOM is about equally effective as non-advertising-influenced WOM: -r. . . Ihe same proportion of consumers say they are going to purchase the brand when the WOM was influenced by advertising (49 percent) as when it was not ^ (49 percent). By another measure, advertisinginfluenced WOM is associated with more .nthusiastic recommendations. Fully 46 percent of all advertising-influenced WOM , _ , . , involves a strong recommendation to buy
or try" the brand versus 39 percent of
.^^U. \\rr^\f

iP^.^.'^.l^.l^ervices The home Sports/recreation/hobbies fi^H^ehold products Health/healthcare Technology Telecommunications Retail Automotive Beverages ^.9..'^!^'}!^}3. Travel services children's products Personal care/beauty ..... ,, ;,.,.,^,...

-MO** -t-10** -t-lO** +9** -i-7** -.6** -1-6** +6** +5** -1-5** +4** -f3 -h3

Other WOM. Meanwhile, WOM that does not involve a reference to advertising is

UM

\A7r\\ L\.. i j

Source: Kell-r Fay Group's TalkTrack*. October 2007 tltroush September 2008

^"^^'^ Brand comiersations influenced by advertising, n = 32.496

Advertismg-reltea WOM means conversations timt contain a reference to advertising as an information source.

"Statistically significant difference at the 90% confidence level.

June 2 0 0 9 M M l OFflOERTISlOGHESEHIICH 1 5 7

ROLE OF ADVERTISING IN WOM

One of the most important discoveries in our research is that 22 percent of aii WOM about brands involve some participant in the conversation referring to something that they saw or heard in paid-media advertising.

GoDiN, SETH. All Marketers Are Liars: The Poiver of Telling Autlientic Stories in a Ixiiv-Trust World. New York: Portfolio, 2005.

GRAHAM, JEFFREY, and

WILLIAM

HAVLEMA.

"Finding the Missing Link: Assessing the Impact of Advertising on Word of Mouth, Web Searches, and Site Visits." Journal of Advertising Research 47, 4 (2007): 427^37.

JAFFE, JOSEPH. Life after the SO-Second Spot: En-

hobbies, and household products. Only for the categories of travel, children's products, and personal care/beauty is the difference not statistically significant.
CONCLUSION

ergize Your Brand with a Bold Mix of Alternatives

ED KELLER IS the CEO of the Keller Fay Group, a market research firm tbat specializes in WOM marketing. He is the co-author of The Infiuentials: One American in Ten Tells the Other Nine How to Vote, Where to Eat, and What to Buy {Free Press. 2003). Mr. Keliec is the past president and the current treasurer of the Board of the Word of Mouth Marketing Association (WOMMA). a board mfimber of the Advertising Research Foundation (ARF), and the past president of the Market Research Council. KELLER, ED, and BRAD FAY. "Leading the ConKATZ, ELIHU, and PAUL F. LAZARSFELD. Per-

to Traditional Advertising. Hoboken, NJ: John Wiley & Sons, 2005.

sonal Influence. New Brunswick, NJ: Transaction Pubiishers, 1955.

Around 20 percent of WOM about brands involves mentioning paid advertising as the source of the information. This suggests a conservative estimate that at least 20 percent of WOM occurs because of advertising. Furthermore, this sort of WOM appears to be of higher quality in that it is more likely to involve a recommendation to buy the brand. This finding supports the "tw^ostep"flowdiscovered 50 years ago, and our results generalize across the 15 different product categories we investigated. These findings raise the question of whether advertising more deliberately designed to stimulate WOM could increase its reach and effectiveness, over more traditional campaigns focused on awareness and persuasion. This is a question that may be best answered with case studies focused on the growing number of advertisers who are making WOM an explicit objective for their campaigns.

versation: Influencers' Impact on Word of Mouth and Brand Conversation." Measuring Word of

BRAD FAY is the chief operating officer of the Keller Fay Group. In 2007, he won the Grand Innovation Award ot the Advertising Research Founcation for the development of Keller Fay's TalkTrack^, the first continuous measurement system for all WOM. including the 90 percent of WOM that occurs offline. Mr. Fay is co-chair of the WOMMA's Influencer Marketing Council.

Mouth, Vol. 3. WOMMA Resarch Anthology, Autumn, 2007.

RiES, AL, and LAURA RIRS. The Fall of Advertis-

ing and the Rise of PR. New York: HarperCollins Publishers, Inc., 2002.

SERNOVITZ, ANDY. Word of Mouth

Marketing:

How Smart Companies Get People Talking. New REFERENCES ADVERTISING AGE. "Family and Friends Most Influential on Shoppers," April 9, 2008. BAYUS, B. L . "Word of MoLith. The Indirect Effects of Marketing Efforts." journal of Advertising Research 25, 3 (1985): 31-39.
ZvMAN, SERGIO, and AKMIN BROTT. Tlie End of

York: Kaplan. 2006.

Advaiising as We Know It. Hoboken, NJ: John Wiley & Sons, 2002.

158

OF HDUEHTISinG RESEIJCH June 2 o o g