Social processes underlying the adoption of products: An omission

Abstract

This paper focuses on one aspect of the social interaction affecting the adoption of products: conversations between product users. When a product is successful, this interaction will increase as users accumulate. These interactions are driven by common interest and can widen users’ knowledge and appreciation of the product. This may raise the volume and persuasiveness of recommendations made to nonusers. Despite this, inter-user conversations are largely ignored in the reported research on the adoption of products.

Introduction

In social science, marketing, and other disciplines, diffusion of innovation is used to describe how new ideas, practices and products (hereafter just products) become adopted. This is an important field because innovations can help solve social and technical problems, and it is expensive when new products fail. For these reasons we need to understand the processes involved in successful diffusion. There is broad agreement on the basic process: that regulation, advertising, conflict and other societal events usually start the adoption of products. Then, for many types of product, the first adopters influence others by word of mouth and example; some of these others adopt the product and then influence yet more people, and so on. This socially driven spread in consumption is often more powerful than any continuing influence from advertising and other societally-based publicity and fits the way a disease spreads (as noted by Tarde, 1903; Rogers, 2003; Gladwell, 2024, and others). Underlying the spread of diseases are processes of contagion and infection and, likewise, the adoption of innovations comes about because of the social interactions between people.

The social interactions that may affect behaviour are observation of others and positive word of mouth. The positive word of mouth about products is mostly produced by those who are using or have used the product; East et al. (2011) found averages of 71% and 22%, respectively, across a sample of categories. Two sorts of word of mouth can be distinguished involving users: that between fellow users of a product, and that between users and nonusers. The inter-user word of mouth is aided by the common interest in the product and, in the case of some products such as transport, sport and entertainment, physical proximity. Observation and positive word of mouth may bring applications of a product to the attention of users and maintain their interest, and this will support continued purchase by users of repetitively-bought products. Inter-user interaction can also provide users with arguments, or scripts, that they may use later with nonusers. In this way, conversations between users may indirectly bring about adoptions. East et al. (2017) found that those who had recently received positive comments about their brand gave nearly twice as many recommendations as those that had not. East et al. proposed that inter-user conversations led to social amplification, making users more willing to pass the idea to others. This evidence seems consistent with everyday experience. Most of us can recall conversations with fellow users that we may later employ in exchanges with both users and nonusers. The problem here is that we do not know what proportion of adoption is generated by social amplification and what proportion by transmission of experience unaffected by any inter-user conversation.

The number of inter-user conversations depends on how many users there are. In some cases, the initial societal impulse will affect many people. For example, a well-publicised change in taxation might be understood by a substantial proportion of the population and could generate wide discussion among these people. By contrast, new goods and services would have to be bought and used before inter-user conversation could take place. On a random basis, we would expect inter-user conversations to be a function of the square of the proportion that have adopted. However, this process will not be random and many other variables are involved.

Factors affecting growth in adoption

The evidence on successful new product adoption has generally shown an initial acceleration in the adoption of a successful product until a point of inflection is reached when the rate of adoption starts to wane: this is the well-known S-shape, first noted by Tarde (1903), elaborated by Rogers (2003), and given mathematical form by Bass (1969. A simple explanation for the S-shape is that adoption based on social influence will depend on the number who have already adopted and the number of persons who could adopt but have not yet done so. Where x is the proportion who have adopted and 1-x is the proportion yet to adopt, this gives a y = x(1-x) function, where y is adoption rate, which has a maximum at x = 0.5 when half the potential users have adopted. This leaves aside any continuing influence of societal factors. In addition, this thinking does not allow for variation in resistance to a new idea which is likely to be larger among those still not adopting. Nor does it allow for loss of interest in the new idea once its acceptance is routine, which could reduce inter-user discussion. There are other forces at work as a new product matures: distribution becomes more established, advertising may be scaled back, products may be improved, competition may develop and prices may fall. In addition, there will be variation in the role of different factors associated with the nature of the product. The previous discussion also casts some doubt on the y = x(1-x) formula since, if the number of inter-user conversations depends on x ² , the formula y = x ² (1-x) would be appropriate for this element of the process; this has a maximum at x = .67.

Investigating the effects of inter-user conversations on adoption

There is work on the effect of customer to customer interaction (Libai et al., 2010) but the mechanisms involved seem unspecified. There is also work on brand communities by Jiang et al. (2023) but this does not define the way in which inter-user conversations could precede adoption. Thus, we have found no direct research of the effect on adoption of inter-user conversations other than that by East et al. (2017). This omission may be partly due to the difficulty of studying real social exchanges that take place infrequently. Here, we consider how this matter might be studied further.

Surveys

Because it is based on surveys, the East et al. study could not demonstrate causation though it is consistent with the claim that inter-user conversations drive adoption. Despite this limitation, some further survey work could show the frequencies and associations of reported inter-user and user-nonuser exchanges when adoption takes place. Such findings could improve the assumptions made when adoption processes are modelled. If surveys are extended by asking initial respondents to answer further questionnaires, some evolution of the adoption process could be illustrated.

Experiments

Experimental designs on word of mouth may limit studies to the effect of single utterances by one person on another. In practice, word-of-mouth exchanges often involve questions and answers and a design that does not permit such discussions is unrealistic. However, Christiansen and Tax (2000) devised a method that is more typical of everyday exchanges and this method could be adapted. They required pairs of participants to each recommend a product of their own choosing to the other, deciding for themselves how they did this. Using this approach to investigate inter-user behaviour, the advising participant would be asked to talk about products that they were using, or had used. The receiving participant would sometimes be another user and sometimes a nonuser. Both participants could engage in conversation about the product, as in everyday exchanges. After the exchange, the experimenter would determine whether recipients were users, ex-users or nonusers, and both participants would be asked how many times they had spoken positively about the product in the previous four weeks. Then, four weeks later, all participants would be asked again how many times they had spoken positively about the product over the previous four weeks. The first set of questions provides base rates and the second, after subtracting the base rate, would show how much users and nonusers had been influenced by the other participant. This would identify the proportions of user:user and user:nonuser interactions and their effects. The procedure would also show whether the advising participant gave more advice as a result of having given advice four weeks earlier. This research design would not be truly experimental because the users and nonusers are not randomly assigned to a product and the repeat questioning of participants may distort their responses, but the method is realistic.

Agent-based models

Agent-based models are also used to represent the hypothesised interactions between people in computer simulations with the outcomes being compared with observed effects. These models are based on assumptions about the volume of word of mouth from adoptees, the success rate of their interactions in gaining adoptions and the social network within which dealings take place. These assumptions are programmed and, when the computer runs, outcomes are illustrated. The effect of different assumptions is shown by varying these assumptions in repeated runs. We have not found models that consider inter-user interaction as an indirect driver of influence on those yet to adopt. This could be incorporated in the starting assumptions and, following our earlier review, the model could test whether a mix of the y = x(1-x) and y = x ² (1-x) functions improves outcomes. A weakness of agent-based models is their reliance on assumptions; these will vary with the category and could benefit from more survey data. Modellers usually include the effect of negative word of mouth and this could be included in survey and experimental work. However, East et al. (2007) found a 3.1:1 ratio of positive to negative word of mouth incidence and, when the product is successful, negative word-of-mouth effects should be even more limited.

Modellers may continue to offer refinements to existing mathematical models but, when so many variables are involved, there is a case for a more empirical approach by basing predictions on the past sales growth patterns for similar products; Wright and Stern (2015) found that this procedure gave strong predictions for categories with low social influence.

Advertising

Advertising is normally directed to users when a product is repetitively bought. This is often described as a form of reinforcement designed to maintain loyalty but it may also have the effect of stimulating conversation between users which may raise the level of advice that they give to nonusers. If so, it may be worthwhile designing advertising to show conversation about the product between users that includes content that could be expressed in word of mouth. This matter is important because a substantial fraction of word of mouth rests on advertising. Keller and Fay (2012) reported that 25% of word of mouth was stimulated by advertising. Lomax and East (2016), comparing small samples of services and durable goods, found that 41% of word of mouth about durables was stimulated by advertising but only 5% service word of mouth rested on advertising. This evidence suggests that it is particularly important to design durable advertising so that it activates word of mouth.