nach oben

Quantitative Marketing and Economics

Erschienen in:

01.09.2009

Targeted advertising as a signal

verfasst von: Bharat N. Anand, Ron Shachar

Erschienen in: Quantitative Marketing and Economics | Ausgabe 3/2009

Einloggen

Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.

search-config

KI-gestützte Suche

Aus

Abstract

This study presents a signaling model of advertising for horizontally differentiated products. The central ingredients of the model are two important characteristics of advertising—targeting, and noisy information content. The theory yields interesting results about the informational role of targeted advertising, and its consequences. First, targeting can itself serve as a signal on product attributes. Second, the effectiveness of targeting depends not only on firms knowing consumer preferences, but on consumers knowing that firms know this. This creates a distinction between strategies of targeting and personalization. Third, the effectiveness of targeting in equilibrium may (far) exceed the information contained directly in the targeted message. Fourth, information content is not, however, superfluous. Specifically, when ads contain no information, a targeting equilibrium does not exist. Together, these results reveal how advertising conveys information both through the content of the message and the firm’s choice of advertising medium. Furthermore, the model is robust to the various critiques of prior work on ads-as-signals: namely, that ad content is irrelevant, ad exposure is unnecessary, and the choice of ads as signals is inherently arbitrary.

Nächster Artikel Communication strategy in partnership selection

Sie haben noch keine Lizenz? Dann Informieren Sie sich jetzt über unsere Produkte:

Springer Professional "Wirtschaft+Technik"

Online-Abonnement

Mit Springer Professional "Wirtschaft+Technik" erhalten Sie Zugriff auf:

über 102.000 Bücher
über 537 Zeitschriften

aus folgenden Fachgebieten:

Automobil + Motoren
Bauwesen + Immobilien
Business IT + Informatik
Elektrotechnik + Elektronik
Energie + Nachhaltigkeit
Finance + Banking
Management + Führung
Marketing + Vertrieb
Maschinenbau + Werkstoffe
Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

Jetzt informieren

Springer Professional "Wirtschaft"

Online-Abonnement

Mit Springer Professional "Wirtschaft" erhalten Sie Zugriff auf:

über 67.000 Bücher
über 340 Zeitschriften

aus folgenden Fachgebieten:

Bauwesen + Immobilien
Business IT + Informatik
Finance + Banking
Management + Führung
Marketing + Vertrieb
Versicherung + Risiko

Jetzt Wissensvorsprung sichern!

Jetzt informieren

Nur mit Berechtigung zugänglich

As Bernheim and Redding (2001) note, “[E]xisting theory provides no basis for selecting one (nondiscriminatory) signal over another. This is viewed as an inherent shortcoming of all signaling-cum-money-burning explanations of economic behavior...Indeed, the choice of the means to burn money is inherently arbitrary. Generally, the question of why one method of money burning is chosen above all others has been left open.”

Relatedly, one often thinks of some of the largest advertisers, like Proctor & Gamble or General Motors, as directing their ad dollars to a “mass audience.” But, even for these firms, their advertising budget invariably comprises spending for many different products that each target a far narrower customer base. For example, General Motors (the world’s largest advertiser) spends more than $50 million each on 12 separate brands that include sports cars, family minivans, and heavy trucks. Proctor & Gamble (the second largest advertiser) spends more than $10 million on 45 separate brands, ranging from specialized pet foods like Iams to anti-dandruff shampoo like Head & Shoulders (Advertising Age, June 27, 2005), each with a relatively distinct target customer set.

In other words, the source of the noise can be in the way the ad was created or in the way it was consumed. Indeed, firms face a difficulty in conveying a lot of information in a 30-second advertisement, and consumers may find it hard to pay attention to each advertisement due to factors such as overload and clutter. For example, The Economist notes that “The average American is now subjected to 3,000 marketing messages every day and could not possibly take them all in.” Even as far back as 1979, Webb and Ray (1979) showed that the more the clutter, the lower are average levels of ad recall.

For example, the correct brand recall of the last commercial seen dropped from 18% in 1965 to 12% in 1974 and 7% in 1981. Advertising textbooks also note that the “major problem facing advertisers today is the difficulty of gaining attention in the face of the increase in advertising clutter” (Batra et al. 1996).

Anand and Shachar (2005) demonstrate this “consumption-deterring” effect of advertising, empirically.

This contrasts with other theories of advertising, for example persuasive advertising, or advertising as complements (Becker and Murphy 1993).

(a) There are other, more technical, differences. For example, repeat purchases are replaced with informative advertising as the source of the single crossing property (that gives the high quality or good match firm a greater incentive to incur the advertising cost in equilibrium). (b) Recently Mayzlin and Shin (2008) presented an interesting variation of “advertising as a signal” in a vertical setting. In their model, it is the content of the ad that signals quality. Specifically, they show that in equilibrium the high quality firm engages in vague image-based advertising, while the medium quality firm engages in attribute-based advertising.

While Bhardwaj et al. (2008) study a different marketing communication tool (salespeople) they are also interested in imperfect communications abilities. In their case, the imperfection is not due to the noisiness of the message but rather to the limited “bandwidth” between the firm and customers.

Chen et al. (2001) discuss the role of both targetability (i.e., firms’ knowledge about consumers’ preferences), and addressability (i.e., the segmentation of media channels) in individual marketing. Their focus, however, is on targeted pricing not on the targeting of ads.

The popularity of Law and Order among Republicans and as a result its attractiveness for the Republican candidate in the 2004 elections are discussed in a New York Times article (Jim Rutenberg, July 18, 2004). While the article also mentions programs that were appealing to the Democratic candidate, none of them was a prime time one. We mention Grey’s Anatomy because based on the discussion in the article this prime time show seems to be appealing to Democrats.

Such data sets are offered, for example, by research firms like the Nielsen Research company, Information Resources, Inc, and Simmons Market Research Bureau.

Notice, though, that if we were studying an asymmetric setting (e.g., two consumers with β = 1 and one consumer with β = − 1) prices may have had a signaling role.

Notice that any $q\neq \frac{1}{2}$ is informative. Here, we consider the natural case of $q>\frac{1}{2}$: in other words, “correct messages are more likely than incorrect messages”.

For example, a consumer who received one message that was m _i,1,1 = 1 would, using Bayes’ rule, form the following expectations about the product attributes: E(x ₁|m _i,1,1 = 1) = (2q − 1) and E(x ₂|m _i,1,1 = 1) = (1 − 2q). Based on this single message, the consumer updates her expected utility from both products. Specifically, since $q>\frac{1}{2}$, E(β ₁ x ₁|m _i,1,1 = 1) > 0 and E(β ₁ x ₂|m _i,1,1 = 1) < 0.

The basic setup of our model is similar to Meurer and Stahl (1994) in the sense that they also focus on advertising for horizontally differentiated products. However, they do not allow for targeting of advertisements (from, say, firm 1 to consumer 1), and assume that advertising is perfectly informative (q = 1).

The hidden assumption behind this profit function is that the difference between any product’s price and marginal cost is 1.

The role of the probability of a “tie” (i.e., ${\rm Pr} (\widetilde{g_{s}}=0|n)$) is similar to the one played by the “pivotal voter” in elections (see, for example, Shachar and Nalebuff 1999). The logic behind this role is that an additional (i.e., marginal) ad would not affect the consumer if g _s is far enough from zero (say g _s = 2). The only way that one ad can change the expected revenue is if it creates a tie or break a tie. For a more formal intuition, please follow the proof of Lemma 8.

This may be argued to be the most reasonable equilibrium, both positively and normatively. Since, in any equilibrium, individuals choose firms whose products best fit their tastes, and the firm’s revenue is the same for any $ n^{\ast }>\underline{n}$, it follows that any $n^{\ast }>\underline{n}$ results in firms incurring higher costs without affecting firms’ revenues or consumer welfare.

The probability of mistakes depends both on the noise in ads and on the number of ads aired. As q increases, ads are less noisy so that the probability of mistakes falls. On the other hand, the number of ads aired by each firm may fall with q (recall that n ^ns is not monotonic in q). It is easy to show, however, that the first effect dominates the second as q increases.

Hertzendorf (1993) revised the model of “advertising as a signal” a-la Nelson to account for the possibility that consumers may see fewer ads than were originally purchased. While he found that ads can still serve as a signal, he also demonstrated that “...under reasonable condition...price and advertising expenditures would never be simultaneously employed to signal quality.”

This version is available from the authors upon request.

(a) Assuming otherwise (e.g. there are multiple time periods within the duration of the game and the consumer can switch between the media channels) would lead to additional significant changes in the model. (b) We get similar results to those obtained below for the following setting: there is a unit mass of consumers of each type; a fraction γ of segment 1 watches media channel 1 and the rest watches media 2; and, the viewing habits of the other segment are the same. Thus, γ of the audience of media 1 is of segment 1 and (1 − γ) is of segment 2.

Consider Gmail, the mail service offered by Google. This service includes a very large memory and no spam. Instead, Google guarantees that commercial mail will be directly linked to the interest of the recipient. To achieve this, the firm scans all the personal mail received by the client and determines her area of interest. The reason this strategy goes beyond simple targeting is that Google ensures that the recipient is aware that the ads were designed for people like her. Targeting, on the other hand, does not guarantee that the recipient knows she is being targeted.

See, for example, Anand and Shachar (2005).

Notice that when n is even, there is a positive probability that the number of “correct” messages equals the number of “incorrect” ones. If n is odd, of course, this event can never occur.

When n is odd, the expected revenue of 1 can increase only if g ₁ = − 1 and the perceived content of the additional ad is 1 (in which case it creates a tie). The probability of the first event is ${\rm Pr} (g_{1}=-1|n)$ and the probability of the second event is q. Since the increase in the expected revenue in such a case is $\frac{1}{2}$ (going from 0 to $\frac{1 }{2}$), the unconditional increase in the expected revenue is $\frac{1}{2} q{\rm Pr} (g_{1}=-1|n)$. Accordingly, it is easy to show that the unconditional decrease in the expected revenue is $\frac{1}{2}(1-q){\rm Pr} (g_{1}=1|n)$. Thus, in order to show that for an odd n an additional ad does not change the expected revenue, we need to show that $q{\rm Pr} (g_{1}=-1|n)=(1-q){\rm Pr} (g_{1}=1|n)$. This is immediate from the fact that for any set of realizations that leads to g _s = − 1 there is a “mirror” set that leads to g _s = − 1 (for example, { − 1,1, − 1} and {1, − 1,1}). It is obvious that the probability of the first set is equal to $\frac{1-q}{q}$ times the probability of the second set.

Notice that if $g_{s^{\prime }}=g^{\ast }$ the consumer is actually indifferent between s and s ^′. However, it is immediate that firm s ^′ will never set a price so that $g_{s^{\prime }}=g^{\ast }$ since by lowering its price by ε the consumer will prefer it.

Let p ^∗(x) be the optimal price given x, and accordingly, one can write A(x) as A(x,p ^∗(x)). Let’s assume that A(x) is a decreasing in x. Thus, $p^{\ast}(x\!+\!1)\{ 1\!+\!{\rm Pr} ( \widetilde{g} _{s^{\prime }}\geq g^{\ast }( p^{\ast }(x\!+\!1),\alpha,q) |2( x +$ $1) ) \} >p^{\ast }(x)\{ 1\!+\!{\rm Pr} ( \widetilde{g} _{s^{\prime }}\geq g^{\ast }\left( p^{\ast }(x),\alpha ,q\right) |2(x)) \} $. However, we know that $p^{\ast }(x+1)\{ 1+ {\rm Pr} ( \widetilde{g}_{s^{\prime }}\geq$ $ g^{\ast }\left( p^{\ast }(x+1),\alpha ,q\right) |2\left( x+1\right) )\} <p^{\ast }(x+1)\left\{ 1+{\rm Pr} \left( \widetilde{g}_{s^{\prime }}\geq g^{\ast }\left( p^{\ast }(x+1),\alpha ,q\right) |2x\right) \right\} $ since ${\rm Pr} ( \widetilde{g}_{s^{\prime }}\geq $ $g^{\ast }\left( p,\alpha ,q\right) |2x) $ decreases in x (since this is the “wrong” firm). Thus, we get that p ^∗(x + 1){ 1 + ${\rm Pr} \left( \widetilde{g}_{s^{\prime }}\geq g^{\ast }\left( p^{\ast }(x+1),\alpha ,q\right) |2x\right) \} > p^{\ast }(x)\left\{ 1+{\rm Pr} \left( \widetilde{g}_{s^{\prime }}\geq g^{\ast }\left( p^{\ast }(x),\alpha,q\right) |2x\right) \right\} $ which means that p ^∗(x) was not optimal.

Abernethy, A., & Franke, G. (1996). The information content of advertising: A meta-analysis. Journal of Advertising, 25(Summer), 1–17.

Anand, B., & Shachar, R. (2005). Advertising, the matchmaker. http://www.tau.ac.il/~rroonn/Papers/match05.pdf.

Anand, B., & Shachar, R. (2007). (Noisy) Communication. Quantitative Marketing and Economics, 5(3), 211–232.CrossRef

Anderson, S., & Renault, R. (2006). Advertising content. American Economic Review, 96(1), 93–113.CrossRef

Bagwell, K., & Ramey, G. (1994). Coordination economies, advertising, and search behavior in retail markets. American Economic Review, 84(3), 498–517.

Batra, R., Myers J. G., & Aaker, D. (1996). Advertising management. Prentice Hall.

Becker, G., & Murphy, K. (1993). A simple theory of advertising as a good or bad. Quarterly Journal of Economics, 108(4), 941–964.CrossRef

Bernheim, D., & Redding, L. (2001). Optimal money burning: Theory and application to corporate dividends. Journal of Economics & Management Strategy, 10(4), 463–507.CrossRef

Bhardwaj, P., Chen, Y., & Godes, D. (2008). Buyer-initiated vs. seller initiated information revelation. Management Science, 54(6), 1104–1114.CrossRef

Butters, G. (1977). Equilibrium distribution of sales and advertising prices. Review of Economic Studies, 44, 465–91.CrossRef

Chen, Y., Narasimhan, C., & Zhang, J. (2001). Individual marketing with imperfect targetability. Marketing Science, 20(1), 23–41.CrossRef

Grossman, G., & Shapiro, C. (1984). Informative advertising with differentiated products. Review of Economic Studies, L1(1), 63–81.

Hertzendorf, M. N. (1993). I’m not a high-quality firm—but I play one on TV. Rand Journal of Economics, 24(2), 236–247.CrossRef

Iyer, G., Soberman, D., & Villas-Boas, M. (2005). The targeting of advertising. Marketing Science, 24(3), 461–476.CrossRef

Jacoby, J., & Hoyer, W. (1982). Viewer miscomprehension of televised communication: Selected findings. Journal of Marketing, 46(4), 12–26.CrossRef

Jacoby, J., & Hoyer, W. (1989). The comprehension/miscomprehension of print communcation: Selected findings. Journal of Consumer Research, 15(4), 434–443.CrossRef

Johnson, J., & Myatt, D. (2006). On the simple economics of advertising, marketing, and product design. American Economic Review, 96(3), 756–784.CrossRef

Kihlstrom, R., & Riordan, M. (1984). Advertising as a signal. Journal of Political Economy, 92(3), 427–450.CrossRef

Mayzlin, D., & Shin, J. (2008). A model of image advertising. Mimeo.

Meurer, M., & Stahl, D. (1994). Informative advertising and product match. International Journal of Industrial Organization, 12(1), 1–19.CrossRef

Milgrom, P., & Roberts, J. (1986). Price and advertising signals of product quality. Journal of Political Economy, 94(4), 796–821.CrossRef

Narasimhan, C. (1988). Competitive promotional strategies. Journal of Business, 61(4), 427–449.CrossRef

Nelson, P. (1974). Advertising as information. Journal of Political Economy, 82(4), 729–754.CrossRef

Resnik, A., & Stern, B. (1977). An analysis of information content in television advertising. Journal of Marketing, 41(1), 50–53.CrossRef

Shachar, R., & Nalebuff, B. (1999). Follow the leader: Theory and evidence on political participation. American Economic Review, 89(3), 525–549.CrossRef

Webb, P. H., & Ray, M. (1979). Effects of TV clutter. Journal of Advertising Research, 9(3), 7–12.

Titel: Targeted advertising as a signal
verfasst von: Bharat N. Anand
Ron Shachar
Publikationsdatum: 01.09.2009
Verlag: Springer US
Erschienen in: Quantitative Marketing and Economics / Ausgabe 3/2009
Print ISSN: 1570-7156
Elektronische ISSN: 1573-711X
DOI: https://doi.org/10.1007/s11129-009-9068-x