Impact of Online Word of Mouth on Channel Disintermediation for Information Goods

Abstract

With the recent advances in digital technology, creators of information goods such as books, songs, or videos can now self-publish and sell their work directly to consumers, without the help of traditional publishers. In this study, we construct an analytical model to examine the role of online word of mouth in this trend of disintermediation. Online word of mouth can reveal product quality for experience goods and raise awareness to accelerate product diffusion. Intuitively, one would expect that with online word of mouth, creators would be more likely to skip publishers and sell directly to consumers. However, our results suggest that online word of mouth may, in fact, encourage more creators to use publishers for high quality work. Our model also makes predictions on the conditions under which online word of mouth benefits publishers and for what types of products and creators it has the most impact.

Key words and phrases:

• Information goods
• content goods
• self-publishing
• channel disintermediation
• online word of mouth
• eWoM

Supplemental Material

Supplemental data for this article can be accessed on the publisher’s website.

Notes

¹ http://media.bowker.com/documents/bowker_selfpublishing_report2013.pdf

² In the week of July 28, 2013, there were four self-published books on the New York Times Bestseller List. Self-publishing author Darcie Chan’s Mill River Recluse hit all three major bestseller lists and appeared on the New York Times bestseller list for 28 weeks [1].

³ Alternatively, we can assume $x_i$ follows a beta distribution with the density function $\frac{x^{\alpha -1}{\left(1-x\right)}^{\beta -1}}{B\left(\alpha ,\beta \right)}$ where $B\left(\alpha ,\beta \right)$ is the beta function. The beta distribution can take different shapes when the values of the two parameters $\alpha$ and $\beta$ vary. Our main results remain qualitatively the same under this alternative assumption.

⁴ Given $t>1$, for any value of $q$ ($0<q<1$) and $p_{s,T}^c$ ($p_{s,T}^c\ge 0$), the expected utility is always negative for the consumers with $x_i=1$, ensuring that the market is never fully covered.

⁵ We have also examined the scenario in which the size of the population is different in the two periods and the scenario in which the consumers who are unaware of the product in the first period may stay in the market and become aware of the product in the second period. Our main results remain qualitatively the same.

⁶ For example, traditional publisher in book industry can buy book reviews, offer book tours, or even give free copies [2, 32]. This is usually hard to achieve by the self-publishing authors due to budget constraints [1].

⁷ This assumption, together with the assumption of $0<f_s<\frac{1}{2}$, simplifies the boundary condition to ensure that $M_{s,T}^c<1$, that is, to ensure that not all consumers are aware of the product. Such assumptions are made to reflect reality and preserve tractability. If we relax these assumptions, we are not able to solve the model in closed form, but our numerical results suggest that our main results hold qualitatively if we assume $R\le 1$ and $f_s<1$ instead.

⁸ In the publishing industry, the author(s) gets a certain proportion of the revenue from sales as author royalty [32, p. 158–160]. In practice, the author’s royalty rate from a traditional publisher usually does not exceed 30 percent [18, p. 52]. However, for the self-publishing platform, the royalty rate can be up to 70 percent to 80 percent [29].

⁹ $\bar{K}=Max\left\{\frac{\left(1-r^{Pb}\right)r^{Pb}}{\left(r^{Sp}-r^{Pb}\right)}\frac{6561}{37120},\frac{81\left(1-r_{Pb}\right)}{256}\right\}$. Detailed derivations of the conditions $k>\frac{\left(1-r^{Pb}\right)r^{Pb}}{\left(r^{Sp}-r^{Pb}\right)}\frac{6561}{37120}$ and $k>\frac{81\left(1-r_{Pb}\right)}{256}$are in the proofs of Lemma 1 and Proposition 3, respectively, in the online Appendix.

¹⁰ Our main results hold qualitatively if only a portion of consumers have access to online WOM to learn about the true quality of the product while the rest of the consumers are unaware of online WOM and keep their expectation as $q{e}_{w,1}=q{e}_{w,2}=\mu$. In a model extension, we also examine the situation in which consumers are only able to estimate the true quality in period 2, but not in period 1, that is, $q{e}_{w,1}=\mu ,q{e}_{w,2}=q$. We analyze the separating equilibrium in which price and advertising become signals of quality and find our main results to hold qualitatively.

¹¹ As we will show later, when there is no online WOM, all the decisions (advertising, pricing and channel choice) are made based on the expected quality $\mu$ and are not reflective of the true quality $q$. Therefore, consumers are not able to use information from advertising, pricing or channel choice to estimate the true quality $q$. This is equivalent to a pooling equilibrium. Separating equilibrium cannot sustain in the case without online WOM because there is not a deterring mechanism here similar to that used in repeat purchase studies (e.g., [41, 59, 63, 68]) to prevent the low-type player from mimicking the high-type player.

¹² We denote $q{e}_{s,1}=q{e}_{s,2}=q{e}_s$ and derive that $\frac{\mathrm{\partial }{R}_s^{\ast }}{\mathrm{\partial }q{e}_s}=\frac{q{e}_s\left(1-r^{Pb}\right)r^{Pb}{\left(f_{s}q{e}_s+4t\right)}^3\left(f_s^{3}q{e}_s^3+10f_s^{2}q{e}_s^{2}t+56f_{s}q{e}_s{t}^2+64t^3\right)}{32k\left(r^{Sp}-r^{Pb}\right)t^3{\left(f_s^{2}q{e}_s^2+8f_{s}q{e}_{s}t+32t^2\right)}^2}$ and $\frac{\mathrm{\partial }{R}_s^{\ast }}{\mathrm{\partial }{f}_s}=\frac{q{e}_s^3\left(1-r^{Pb}\right)r^{Pb}{\left(f_{s}q{e}_s+4t\right)}^3\left(f_s^{2}q{e}_s^2+8f_{s}q{e}_{s}t+48t^2\right)}{64k\left(r^{Sp}-r^{Pb}\right)t^3{\left(f_s^{2}q{e}_s^2+8f_{s}q{e}_{s}t+32t^2\right)}^2}.$ Both derivatives are greater than 0.

¹³ Note that when there is no online WOM, consumers are not able to infer quality from creators’ channel choice because their channel choice depends on consumers’ expected quality $\mu$ (i.e., the mean of the distribution $F$), but not the true quality $q$. When there is online WOM, consumers can observe $q$ from online WOM and thus do not need to infer quality from the creators’ channel choice.

¹⁴ When $R>\frac{\left(1-r^{Pb}\right)r^{Pb}{\left(f_s+4t\right)}^4}{128k\left(r^{Sp}-r^{Pb}\right)t^3\left(f_s^2+8f_{s}t+32t^2\right)}$, for any $q{e}_s$ ($0<q{e}_s<1$), we always have $R_s^{\ast }<R$, that is, the creator will always choose to self-publish.

¹⁵ The detailed expression of $\hat{q}$ is too complex to provide, but we prove its existence in the proof of Proposition 4 in the online Appendix.

¹⁶ Note that a creator’s channel decision is based on consumers’ expected quality under each scenario. Specifically, under scenario o, this decision is based solely on μ, whereas q does not play any role in this decision. Under scenario w, this decision does depend on q but q is revealed by online WOM. When we examine the effect of online WOM on a creator’s channel decision, q is again revealed by online WOM. Therefore, information asymmetry does not play a role in the creator’s channel decision here, and thus the question of whether channel decision can signal product quality does not apply.

¹⁷ We use $\mu =0.7,a=0.1,\mathrm{\Delta }a=0.1,t=1,k=0.3,r^{Pb}=0.3\mathrm{a}\mathrm{n}\mathrm{d}{r}^{Sp}=0.7$ for all the figures (unless otherwise specified) as an illustrative example. Our results remain qualitatively the same if other parameters are used.

¹⁸ http://www.thedailybeast.com/articles/2011/03/24/barry-eisler-explains-self-publishing-decision.html. http://www.forbes.com/sites/juliapimsleur/2014/11/04/traditional-publishing-vs-self-publishing.

¹⁹ http://www.hockingbooks.com/how-i-became-a-published-author/.

²⁰ https://www.reportbuyer.com/product/4442240/consumer-books-global-market-briefing-2017.html

²¹ http://www.forbes.com/sites/jaymcgregor/2015/04/17/mark-dawson-made-750000-from-self-published-amazon-books

http://www.forbes.com/sites/jaymcgregor/2015/04/17/mark-dawson-made-750000-from-self-published-amazon-books/2

Additional information

Notes on contributors

Ho Cheung Brian Lee

Ho Cheung Brian Lee ([email protected]; corresponding author) is an Assistant Professor of Operations and Information Systems, Manning School of Business, University of Massachusetts Lowell. He obtained his Ph.D. from the School of Business, University of Connecticut. Dr. Lee’s primary research interests include the economics of information systems, crowdsourcing, and platform strategies. His work has appeared in Information Systems Research and Decision Sciences.

Xinxin Li

Xinxin Li ([email protected]) is an Associate Professor of Operations and Information Management at the School of Business, University of Connecticut. She received her Ph.D. from The Wharton School at University of Pennsylvania. Dr. Li’s research interests lie at the intersection of information systems and marketing, with an emphasis on the implications of new technologies to consumer welfare, firm pricing and competitive strategies. Her work has been published in Management Science, MIS Quarterly, Information Systems Research, and Marketing Science, among other journals. She is an Associate Editor for MIS Quarterly.