The social value of word of mouth programs
We investigate how acceleration and acquisition combine to create value in a word-of-mouth seeding program for a new product. We define the social value of a program as the global change, over the entire social system, in customer equity that can be attributed to the word of mouth of program participants. We compute the social value of programs in various scenarios, using empirical connectivity data on 12 social networks in various markets as an input for an agent-based model that simulates the diffusion of a new product in a competitive scenario.
We show how the presence of competition, program size, and choice of program members (randomly selected or selected from the influentials) affect the social value and the relative contributions of acceleration and acquisition to this value. .
Power of hubs - analytics
Aggregating individual-level social network considerations to the market level, we demonstrate that seeding hubs has a differential accelerating effect on diffusion measured by the additional net present value (NPV) of potential future sales. On the basis of closed-form solutions, we find that where consumers’ decisions to purchase a new product are almost entirely induced by word-of-mouth communications, seeding a small number of hubs whose social-connectedness is about 10 times greater than that of ordinary individuals, may help initiate a valuable diffusion process in which the NPV is increased by several tens of percentage points. On the other hand, seeding such highly connected hubs adds less than 1% to the NPV. Tapping into a category of social influence that is characterized by the number and intensity of social ties, we find that a hub’s “area of influence” has greater impact on NPV than its tie intensity.
Power of hubs - empirical evidence
We explore the role of hubs (people with an exceptionally large number of social ties) in diffusion and adoption. Using data on a large network with multiple adoptions, we identify two types of hubs—innovative and follower hubs. Contrary to recent arguments, hubs tend to adopt earlier in the diffusion process, even though they are not necessarily innovative. Although innovative hubs have a greater impact on the speed of the adoption process, follower hubs have a greater impact on market size (total number of adoptions). Importantly, a small sample of hubs offers accurate success versus failure predictions early in the diffusion process.
Zooming in: Finding clues in microdata
By analyzing the high volatility of daily data, we show how changes in sales patterns can self-emerge as a direct consequence of the stochastic nature of the process. Our contention is that the fluctuations observed in more granular data are not noise, but rather consist of accurate measurement and contain valuable information. By stepping into the noise-like data and treating it as information, we generated better short-term predictions even at very early stages of the penetration process. Using a Kalman-Filter-based tracker, we demonstrate how movements can be traced and how predictions can be significantly improved. We propose that for such tasks, daily data offer more insights than do smoothed annual data.
Agent-based models and differential equations
One of the most prominent models in diffusion theory is the Bass Diffusion model that describes the number of adopters of new products as a differential equation. More recently, diffusion of products has been studied using agent-based modeling and in particular, stochastic cellular automata models. In this study we explore the relation between the Bass model and the cellular automata models. In particular, we explore the differences between the two models that is due to the fact that time is continuous in the Bass model but discrete in cellular automata models, and describe the conditions under which the cellular automata models converge to the Bass model. Empirical verifications of the main results are also discussed.
The power of seeding in multi-market entry
Managers introducing new products into multi-markets have to respond to varying regional developments over time. Firms may choose from among three resource allocation strategies: uniform strategy, where the firm distributes its marketing efforts evenly among its regions; support-the-strong strategy, where the firm invests its efforts proportional to the percentage of adopters in that region (at least up to a certain market coverage); and support-the-weak strategy, where the firm invests its efforts proportional to the remaining region potential. We find that strategies which disperse marketing efforts, such as support-the-weak and uniform strategies are generally superior to support-the-strong strategy. Not only is this finding surprisingly robust to different market conditions and variations on these strategies, but it also runs counter to conventional wisdom commonly found in research and business press on international marketing.
Using spatial data for early prediction
One of the main problems of early-period assessment of new product success is lack of sufficient data to enable reliable predictions. We show that managers can use the spatial dimension of sales data to obtain predictive assessment of the success of a new product shortly after launch time. Relying on diffusion theory, we expect that the success of many innovative products highly depends on word-of-mouth communications. Because word-of-mouth spread is often associated with geographical proximity between the parties involved, one can expect that "clusters" of adopters will begin to form. Accordingly, we can examine how "far" the process is from a uniform geographical distribution: A product whose distribution is further away from a uniform distribution will have a higher likelihood of beginning a "contagion process" and therefore of being a success.
Resistance to innovations
Limited attention has so far been given to the adverse effect of negative word of mouth and consumers’ resistance to change. In this paper, we explore how resistance may shrink market size. In light of the covert nature of negative word of mouth, we use cellular automata modeling to simulate and gain insights into possible resistance scenarios and their implications. We found that advertising provides no more than a limited compensating effect, and positive opinion leaders have low impact on market growth. We explore an approach that undermines the effect of resistance leaders through the direct activation of positive opinion leaders prior to the initiation of unfocused marketing efforts.
Network effects on the diffusion of innovations
We study the influence of network topology on the speed and reach of new product diffusion. We study simultaneously the effect of three network metrics: the average degree, the relative degree of social hubs (i.e., the ratio of the average degree of highly-connected individuals to the average degree of the entire population), and the clustering coefficient. A novel network-generation procedure based on random graphs with a planted partition is used to generate 160 networks with a wide range of values for these topological metrics. Using an agent-based model, we simulate diffusion on these networks and check the dependence of the net present value (NPV) of the number of adopters over time on the network metrics. We find that the average degree and the relative degree of social hubs have a positive influence on diffusion.
Targeting Revenue Leaders
Historically, when targeting potential adopters of a new product, firms have tended to focus first on people with disproportional effect on others, often labeled “opinion leaders.” We highlight the benefit of targeting customers with high lifetime value (CLV), or “revenue leaders.” We further argue that targeting revenue leaders can create high value both by accelerating adoption among these customers and because of the higher-thanaverage value that revenue leaders generate by affecting other customers with similarly high CLV. The latter phenomenon is driven by network assortativity, whereby people’s social networks tend to be composed of others
who are similar to themselves.
Uncovering network structure using penetration data
We propose a method for uncovering the structure of the adopters' network underlying the diffusion process, based on penetration data alone. By uncovering the traces that this network leaves on the dissemination process, the degree distribution of the network can be estimated. We show that the network's degree distribution has a significant effect on the contagion properties. Ignoring the network structure introduces significant errors to estimated diffusion parameters and may lead to flawed assessments of the magnitude of the contagion process. In three studies we validate the proposed method using data for known mapped networks and the adoption process propagating on them.
The chilling effects of network externalities
We show that network externalities create a slowdown on growth initially, because potential customers wait for other adopters to provide them with more utility before they adopt. We explore the financial implications of network externalities, taking the entire process into account. Using an agent-based as well as an aggregate-level model, and separating network effects from word of mouth, we find that network externalities have a substantial chilling effect on the Net Present Value associated with new products. Drawing on collective action literature to relate network effects to individual threshold levels, we find that the chilling effect is stronger with a small variability in the threshold distribution, and especially affected by the process early on in the product life cycle. We also show the “hockey-stick” growth pattern, empirically examining the growth of fax machines, CB radios, CD players, DVD players, and cellular services.
The NPV of bad news
While both the business press and academic literature have drawn attention to the possible destructive nature of negative word of mouth on the individual level, little is known about the way individual-level factors aggregate to market-level results. We explore the effect of individual- and network-level negative word of mouth on the firm’s profits using an extended model of small world analysis.We simulate a market in which information spreads when consumers interact with each other using both permanent strong ties within their own social system, and changing, often random, weak ties with other networks. We tie individual-level parameters to aggregate results using regressions as well as structural equations modeling. The effect of negative word of mouth on the net-present-value of the firm is found to be substantial, even when the initial number of dissatisfied customers is relatively small.
Immunization of networks against viruses
Although computer viruses cause tremendous economic loss, defense mechanisms fail to adapt to their rapid evolution. Previous immunization strategies have been characterized as being static and centralized, which has made virus containment difficult or even impossible. We suggest, instead, to propagate the immunization agent as an epidemic. The main problem with epidemic vaccine propagation is that it is bound to lag behind the virus. We suggest giving the vaccine an advantage over the virus by allowing it to leapfrog through a separate, overlapping, partially correlated network. This enables the antivirus to contain the epidemic efficiently. We systemize this concept with a ‘honey-pot’ architecture that achieves both early virus discovery and rapid antivirus dissemination. We present analytic, as well as simulation, results for a set of realistic topologies that illustrate the effectiveness of this approach
Weak cross-market communications create a saddle
Using data on a large number of innovative products in the consumer electronics industry, we find that between one third and one half of the sales cases involved the following pattern: an initial peak, giving rise to a trough of sufficient depth and duration to exclude random fluctuations, followed by sales which eventually exceeded the initial peak. This newly identified pattern, which we call a saddle, is explained by the dual-market phenomenon that treats the early market adopters and main market adopters as sufficiently different to warrant differential treatment as two separate markets for marketing purposes. If the early market and the main market adopt at different rates, and if the communications between these two segments are relatively weak, then overall sales to the two markets will exhibit a temporary decline at an intermediate stage.
The strength of weak ties
There is scant evidence on the breakdown of the personal communication between closer and stronger communications that are within an individual's own personal group (strong ties), and weaker and less personal communications that an individual makes with a wide, often random, set of other acquaintances and colleagues (weak ties). We model these two phenomena and show that the influence of weak ties is at least as strong as the influence of strong ties. Despite the relative inferiority of the weak tie parameter in the model's assumptions (strong ties reflect greater probability for an individual-level transformation), their effect approximates or exceeds that of strong ties, in all of the process stages.
Does heterogeneity matters in innovation diffusion?
Aggregate-level simulation procedures have been used in many areas of marketing. In this paper we show how individual-level simulations may be used to support marketing theory development. More specifically, we address two major issues facing current theories of innovation diffusion: The first is a general lack of data at the individual level, while the second is the resultant inability of marketing researchers to empirically validate the main assumptions used in the aggregate models of innovation diffusion. We show that notwithstanding some exceptions, the well-known Bass model performs well on aggregate data when the assumption that that all adopters have an equal effect on all other potential adopters is relaxed.