aviralbranchingmodelforpredictingthe ...· aviralbranchingmodelforpredictingthe ......
Post on 23-Feb-2019
Embed Size (px)
Vol. 29, No. 2, MarchApril 2010, pp. 348365issn 0732-2399 eissn 1526-548X 10 2902 0348
informs doi 10.1287/mksc.1090.0520
A Viral Branching Model for Predicting theSpread of Electronic Word of Mouth
Ralf van der Lans, Gerrit van BruggenRotterdam School of Management, Erasmus University, 3000 DR Rotterdam, The Netherlands
Jehoshua EliashbergThe Wharton School of the University of Pennsylvania, Philadelphia, Pennsylvania 19104,
Berend WierengaRotterdam School of Management, Erasmus University, 3000 DR Rotterdam, The Netherlands,
In a viral marketing campaign, an organization develops a marketing message and encourages customersto forward this message to their contacts. Despite its increasing popularity, there are no models yet thathelp marketers to predict how many customers a viral marketing campaign will reach and how marketerscan influence this process through marketing activities. This paper develops such a model using the theory ofbranching processes. The proposed viral branching model allows customers to participate in a viral marketingcampaign by (1) opening a seeding e-mail from the organization, (2) opening a viral e-mail from a friend, and(3) responding to other marketing activities such as banners and offline advertising. The model parametersare estimated using individual-level data that become available in large quantities in the early stages of viralmarketing campaigns. The viral branching model is applied to an actual viral marketing campaign in whichover 200,000 customers participated during a six-week period. The results show that the model quickly predictsthe actual reach of the campaign. In addition, the model proves to be a valuable tool to evaluate alternativewhat-if scenarios.
Key words : branching processes; forecasting; Markov processes; online marketing; viral marketing; word ofmouth
History : Received: March 4, 2008; accepted: May 27, 2009; processed by Carl Mela. Published online in Articlesin Advance August 26, 2009.
1. IntroductionIn October 2006, Unilever launched a 75-second viralvideo film, Dove Evolution. This campaign generatedover 2.3 million views in its first 10 days and threetimes more traffic to its website than the 30-secondcommercial that aired during the Super Bowl (vanWyck 2007). More recently, Comic Relief, a Britishcharity organization, achieved 1.16 million partici-pants in the first week after launching their viralgame Let It Flow that promoted Red Nose Day,their main money-raising event (New Media Age 2007).These two examples illustrate a new way of market-ing communication in which organizations encouragecustomers to send e-mails to friends containing a mar-keting message or a link to a commercial website.Because information spreads rapidly on the Internet,viral marketing campaigns have the potential to reachlarge numbers of customers in a short period of time.Not surprisingly, many companies such as Microsoft,Philips, Sony, Ford, BMW, and Procter & Gamble
have gone viral. However, not all viral marketingcampaigns are successful, and because of competi-tive clutter, they need to become increasingly sophis-ticated in order to be effective and successful. It isalso important that marketers be able to predict thereturns on their expenditures and thus how manycustomers they will reach. As one marketing agencyexecutive stated: The move to bring a measure ofpredictability to the still-unpredictable world of viralmarketing is being driven by clients trying to balancethe risks inherent in a new marketing medium withthe need to prove return on investment (Morrissey2007, p. 12). Despite their importance, no forecastingtools for these purposes are available yet. The aim ofthis research is to develop a model that predicts howmany customers a viral marketing campaign reaches,how this reach evolves, and how it depends on mar-keting activities.The structure of this paper is as follows. Section 2
defines viral marketing campaigns and describes
van der Lans et al.: A Viral Branching Model for Predicting the Spread of Electronic Word of MouthMarketing Science 29(2), pp. 348365, 2010 INFORMS 349
how marketers can influence the viral process. Sec-tion 3 shows how the flow of communication amongcustomers in viral marketing campaigns follows abranching process, and we introduce our viral branch-ing model (VBM). Section 4 describes the data of areal-life viral marketing campaign that reached over200,000 customers after only six weeks. The predic-tive performance of our model, analyzed using datafrom this campaign, is presented in 5. The final sec-tion discusses implications of our research and sug-gestions for further research.
2. Viral Marketing CampaignsIn a viral marketing campaign, an organization devel-ops an online marketing message and stimulates cus-tomers to forward this message to members of theirsocial network. These contacts are subsequently moti-vated to forward the message to their contacts, and soon. Because messages from friends are likely to havemore impact than advertising and because informa-tion spreads rapidly over the Internet, viral market-ing is a powerful marketing communication tool thatmay reach many customers in a short period of time(De Bruyn and Lilien 2008). Furthermore, the natureof the Internet allows marketers to use many differ-ent forms of communication such as videos, games,and interactive websites in their viral campaigns. Theterm viral marketing may (incorrectly) suggest thatinformation spreads automatically (Watts and Peretti2007). However, marketers need to actively managethe viral process to facilitate the spread of information(Kalyanam et al. 2007).
2.1. Marketing Activities for Managing ViralMarketing Campaigns
In viral marketing campaigns, marketers may use twotypes of strategies to influence the spread of infor-mation. The first focuses on motivating customers toforward marketing messages to their contacts (Chiuet al. 2007, Godes et al. 2005, Phelps et al. 2004).As suggested by Godes et al. (2005) motivations toforward messages are either intrinsic or extrinsic. Theformer can be triggered by the content of the mar-keting message. Important components of the mar-keting message are the subject line of the e-mail andthe text in the e-mail itself (Bonfrer and Drze 2009).Furthermore, marketers nowadays develop websitescontaining videos and games that attract customerattention and interests. These websites usually facili-tate the viral process by providing tools to easily for-ward e-mails to friends, such as Tell a Friend orShare Video buttons. Examples of extrinsic motiva-tions to forward marketing messages are prizes andother monetary incentives (Biyalogorsky et al. 2001).Although increasing customers motivation to for-
ward messages to friends has a strong impact on the
reach of the viral campaign, this is usually a diffi-cult and expensive task. In contrast, controlling thenumber of initial or seeded customers is much morecost effective. In general, marketers can choose fromthree distinct categories to seed their viral marketingcampaign: (1) seeding e-mails, (2) online advertising,and (3) offline advertising. Seeding e-mails are usu-ally sent by the company itself or by a specializedmarketing agency to customers who have given per-mission to receive promotional e-mails (Bonfrer andDrze 2009). Using this seeding tool, a marketer cantarget a specific group of customers that are poten-tially interested in the campaign. The design and con-tent of the e-mails are crucial because customers easilycategorize such e-mails as spam and quickly deletethem. For this reason, seeding e-mails are expectedto be less effective than viral e-mails that are sent byfriends or acquaintances of the recipient.Online advertising is another important seeding
tool that marketers can use to influence the viralprocess. The effectiveness of online advertising maydiffer depending on the customers as well as the web-sites on which the ads are placed. It is worth notingthat marketers can directly observe when a specificonline ad generates a visitor to the viral campaign.Hence, the effectiveness of online advertising can bemonitored accurately, and based on its performance,marketers can decide to adapt their online advertis-ing strategy. Furthermore, online advertising agenciesoffer contracts that guarantee a predetermined num-ber of clicks to the campaign website within a certaintime window. In such cases organizations usually payfor each click. Because online ads may be perceived asless obtrusive than promotional e-mails, this seedingtool may be very attractive.Finally, besides online seeding tools, marketers may
still use traditional offline advertising to seed theircampaigns. Examples are magazine or TV ads thatrefer to the website of the viral marketing campaign,and package labels or coupons that try to attract vis-itors to the campaign website. However, offline seed-ing is less popular and expected to be less effectivebecause customers cannot directly visit the campaignwebsite by clicking a link. Another disadvantage ofoffline seeding is that it is more difficult to measureits effectiveness, because marketers cannot directlyobserve when offline advertising generates a cus-tomer to the viral campaign. Possible solutions forthis problem are asking customers on the website howthey were informed or to ask for the barcode of theproduct or coupon that was used to enter the website.As described above, the appropriate strategic deci-
sion of the marketing activities at the right momentstrongly depends on the spread of the process andthe effectiveness of each marketing communicationtool. Therefore, marketers need to closely monitor thespread of information in viral marketing