Understanding the role of social influence in consumer behavior

You made me do it

Editor’s note: Jon Christiansen is the director of analytics at Sparks Research, Clemson, S.C.

It’s Thanksgiving morning. The Packers are playing the Lions in Detroit. On this rare occasion, I am spending Thanksgiving with my sister in Ann Arbor, Mich. That morning, waiting for kickoff, I sit back and observe a simple social exchange lead to a possible outcome of mammoth proportions.

My niece, Linnea, was presented with a conflict of choice from her father: Who she is pulling for? The Packers (daddy’s team) or the Lions (mommy’s team)? (Before I continue, I should note that my brother-in-law is a great father and husband and rarely encourages a choice opposite that of my sister. But when the Lions and the Packers play, it is a house divided.) Linnea politely replies, “The Lions, because ‘Lions’ starts with an ‘L,’ like ‘Linnea’.”

A simple, yet rational decision. When offered a choice of high complexity (loyalty to mom vs. loyalty to dad), like most of us, she simplifies her decision process by deflecting to a focal parameter. But then it got interesting; my brother-in-law threw her a curveball: “Well, you know that Michaela [the much-admired older cousin] is a Packers fan, right? So what would you say now?”

Draw conclusions from patterns

A common goal of marketing research is to study the attitudes, perceptions, preferences and behaviors of markets. Our hope as researchers is to draw conclusions from patterns, framed by what these consumers are telling us. This is not a simple task, because research most commonly focuses on measuring marketing-mix variables.

Mix variables include but are not limited to price, consumer awareness, advertising reach, mode of communication, product specs and specs of competitor products, channel distribution and overall brand health. And since tracking these metrics and parameters isn’t enough of a challenge, tack on our humble attempt at personifying consumers and the ever-difficult task of measuring consumer preferences (who was it that said there was no accounting for taste?). And this is just one side of the fence!

It is difficult to tell the whole story from one side of the fence. Unfortunately, the world does not function in static patterns; rather, we live in a dynamic place, one altered by the boundless influences around us. We think twice about visiting a restaurant when a close friend or coworker speaks of a bad experience. We celebrate when a highly-critical friend humbly admits to liking a movie we recommend. We are honored when a family member calls to ask for advice about a major life decision. We are flattered when a friend with markedly higher intelligence needs our “expert opinion.” In summary, we are all the contributors and beneficiaries to a world of dynamic influence, most aptly driven by exchanges between individuals.

No longer handcuffed

In today’s conditions, when it comes to making buying decisions, consumers are smarter, more resourceful and better prepared than ever. Consumers are no longer handcuffed by information asymmetries. Anyone with an Internet connection can research any product they want, whenever they want, from a multitude of sources. Once they make a purchase, they can relay their opinion to many others within minutes. That opinion is assessed by complete strangers on an anonymous message board or by close friends and family members on . And with the wealth of knowledge sources available to consumers, they are commonly inclined to favor information obtained from a focal source. That common focal source: other individuals.

On the other side of the fence is where the exchange between consumers occurs. So we ask, “How important are these exchanges?” To help explain, let’s borrow from one of the most respected product adoption models in existence: the Bass model. The standard Bass model implies that, on average, the coefficient of the social influence and contagion parameter is nearly 13 times greater than that of the function of advertising (and with little deviation from this ratio on average). Using this model, informal roles of generating awareness and subsequent adoption have an exponentially greater impact than that of advertising efforts.

This is not to suggest that advertising is arbitrary but can we blame ourselves for allowing our social exchanges to supplant advertising as our go-to source of influence? The difficulty with advertising is that it rarely discloses the whole story. Rational marketers rarely advertise something negative about their product and we cannot exactly ask questions to people on TV trying to sell us on their product (maybe one day!). Thus we, as consumers, gauge the truth about products from our social exchanges.

But as researchers, this is a unique challenge. At this juncture, researchers have to ask how they can accurately measure what is going on in social circles. How can they understand what sources of influence truly affect product adoption?

Scholars of social influence suggest that we can learn much about consumer behavior by studying the connections in their lives: who or what they have access to, who or what they choose to access, etc. We can learn much about the consumer by understanding what connects them. To simplify, social influence is driven by two primary agents: sources we perceive to think like us (bonding) and sources we perceive to know more than us (the trusted expert). Obviously there are multiple dimensions to understanding this at a deeper level but the core remains consistent.

Much like us

Let’s first explore the bonding function. Whether we notice or not, we tend to surround ourselves with people who are much like us. Our ties likely have similar beliefs, education and socioeconomic status. There is a good chance we went to the same school, are fans of the same teams, dress similarly and have similar hobbies. We naturally surround ourselves with people who have similar needs. Thus, when we see someone close to us make a buying decision, it is safe to say that we believe it would also meet our needs. When we watch someone with whom we share this bond make a purchasing decision, we subconsciously program that product into our choice set. Our choice set has now expanded and our preference for that product is subconsciously climbing up our preference ladder, equipping itself for when it is time to decide.

Consider the following story:

Jack – a chiropractor, family man and part-time volunteer basketball coach – attends a football game with three of his best friends. It is a rare occasion for Jack to spend time with his friends, yet the topics of conversation remain unchanged: football, career, family life, reliving stories from college and the argument over which friend has the coolest toy collection.

Jack realizes quickly that he is losing the arms race for cool toys. This is especially evident to Jack when he notes that all three of his friends are practically tethered to their smartphones, so much that it seemed as if a fifth passenger was along for the ride. This passenger was a know-it-all and Jack rarely cares for know-it-alls. This passenger could squash any argument quickly with its wealth of knowledge.

Jack had never felt a need for a smartphone – his flip phone makes calls, stores numbers, texts occasionally – until he comes to understand what he has been missing this whole time. It wasn’t long before his friends began ragging on him for his antiquated cell phone (you know, the base model flip phone that was discontinued two weeks after he bought it), especially when he wasn’t the guy getting all the score updates, seeing the instant replays from his drive-in movie-sized screen or being the first to answer the random questions, like, “Where did number 94 go to college?” After getting to know this fifth passenger, Jack is converted – this know-it-all is pretty cool after all.

This is not to suggest that friends, family or coworkers are the only sources of the bond exchange. The same can be said for figures we connect with in the media and in sports and for role players in advertising and other avenues of our lives. If we believe in the idea that we share a like mind with someone or something, we have a connection. We might take advice from a columnist because they post on a particular blog that we enjoy. We might be swayed toward certain products because a store we like carries them.

Many faces

Next, consider the role of the expert. We are all experts in some way; some more so than others; some knowledge more valuable than others. The function of the expert is more than the definition implies; there are many faces of the expert. The simplified version of the expert is the source who we perceive has more knowledge than we do about something that interests us. This could be a sales representative or a subject-matter expert. There are sources who individuals admire or even idolize. Then there are those who have an extensive knowledge in the area of interest. While the aforementioned sources have multiple faces, others have no face at all. They might have, at best, a self-proclaimed alias, a handle (e.g., fanguy6317).

Additionally, the level of influence expertise plays expands when knowledge is a product of experience. In this case, I speak of the customer experience. We avoid restaurants with poor recommendations; we consider vacationing in a new destination when someone had a great experience and we share our experiences with others, unknowingly influencing those with whom we share our experiences. Oftentimes, these are friends, family members or coworkers, yet at other times, these are complete strangers. Those of us who have shopped online can likely attest to having fallen victim to the “star ratings” and the comments supporting their rating. Ironically, we in the marketing research field would balk at recommending a business decision with support from a sample size of five but it certainly plays a role when shopping Web sites that use a rating system.

The point is, whether the expert is more knowledgeable or not, the fact that we perceive them to be makes them influential.

Accurately measure

So, I return to my original questions: How can researchers accurately measure what is going on in social circles? How can researchers really understand what sources of influence truly affect product adoption?

The answer is not as clear as we would like. Rather, it is complex, if not improbable to reach a pinpoint estimate. Therefore, researchers are charged with simplifying this puzzle. Two key methods are common in doing so.

Modeling with existing or historical data. The first method is to model product diffusion using existing or historic data of the same or similar products. Product sales data is freely available through a multitude of sources if no in-house data fits. It is best to include a number of possible products and bridge patterns in sales trends, removing those that do not fit appropriately. Using the common elements from the remaining products, a researcher can use this new model to forecast sales for the new product. It is best to keep the model simple if the model will allow for it.

Borrowing from a research example, a colleague of mine and I recently studied the adoption patterns of Google Apps for Education at a mid-size public university. Several years ago, Google developed a platform for colleges and universities to improve e-mail services and faculty student collaboration. Each student who adopted the service would get a new e-mail address (student@t.quirkuniversity.edu) which would merge their existing e-mail account with a customized Google Gmail platform. When we began our research, we utilized nearly 100 estimators in our projected model of adoption, only to discover that diffusion patterns were best explained by only three to four estimators.

Structured analogy forecasting. The second method is more challenging; without existing or historical data, researchers must rely on alternative sources that on the surface appear less reliable. Estimates are most commonly derived from judgments of field experts or company executives, through prediction markets or through simulating adoption patterns of similar products. Focusing on model development, using structured analogy forecasting, it is best to work backward from models that share similar elements to the one in question. In fact, it is surprising how closely the adoption patterns fit when matched with analogous products. A word of caution however – it is important to control for situational criteria. Some examples include: market conditions, availability of product substitutes, supply chain patterns and variation in marketing strategy.

Returning to the Google Apps for Education example, my colleague and I thought it would be interesting to model a projected adoption pattern before the service was deployed. In doing so, we explored products or ideas that would most likely share similar parameters.

Since we were made aware that the administration would place little emphasis on publicizing Google Apps for Education, we determined that social exchange would be the dominant driver of adoption. We also knew that students were well aware of the benefits of Google Apps (especially Gmail). Therefore, we shrunk our search criteria for similar products or concepts to two criteria. First, we looked for products with two key criteria: products or concepts that had a similar ratio of social influence to advertising influence (favoring heavily on the social side). Second, we looked for products or concepts that had some previous market presence. We chose some interesting products and concepts in our model, all of which closely fit each of the two criteria as well as some other more finite criteria (some closer than others, obviously). Our examples included: Facebook, the eight-bit microprocessor, accelerated education programs and the second-generation cell phone. We added a few other educational innovations and found that, after normalizing time to adopt, our forecasted (as well as actual) adoption patterns fit nicely among these analogous products.

Pair what they know

And despite the uncertainty that comes with forecasting product adoption, researchers can pair what they know about consumer behavior with the product makeup. Having explored the behavioral patterns of consumers, here are a few questions to keep in mind when estimating the probable role of social influence:

How likely will consumers see other consumers use the product? Products that are highly visible in use allow for higher propensity of bond influence and the expert influence of the experienced consumer. In this case, the bond influence is often stronger: The mere repetition of observation is a strong enough influence in and of itself.

How valuable is the experience to the product? If the experience makes or breaks product satisfaction, the first few simulation periods are often critical.

How complex is the product? This can help frame the true value of the product expert. If products on the market are less complex, leveraging the product expert can influence those with lesser understanding.

How many online sources are available for product recommendation or criticism? The most easily accessible source of customer experience is online. The online recommendation market is perhaps the key influencer, especially when there are a great number of product substitutes. Not only can consumers gauge the total customer experience, but there is a strong chance we will connect with someone who values the same experience.

Is there a level of social sensitivity to your product? Consumers are most likely to discuss products of social sensitivity (health, financial, etc.) with trusted sources.

Is it a product that most people wouldn’t talk about? If so, discussion will likely be less about experience and more about product attributes and features. These are likely to be bond influences since discussion about product features often occurs between consumers who have similar levels of knowledge and interest.

Is the product segment-specific? Segments are most likely aligned by bonding patterns, since segments are often linked by similar behavioral patterns.

Is the product a function of market conditions? In other words, does the product thrive only in certain conditions? If so, it will likely create contagion (most likely bond-induced) quickly and grow if it adapts with the market cycles.

Is location a factor? If the product is specific to location (or at least starts there), the bonding element is more often the driving force of product adoption.

Is the product cool? In other words, do consumers make a statement about themselves when they adopt the product? This suggests high-level bond effects but also high-level effects from the admired expert.

Growing field of research

Should a researcher desire to test the possible outcomes given the alteration of social influence parameters, a growing field of research is available. Computational modeling (or multi-agent modeling) is a cost-efficient and sound method for simulating conceptual data. The key to computational modeling, similar to other examples listed, is to keep the model simple. In doing so, researchers can experiment with different parameters of the market and the diffusion of social exchanges (examples include the questions outlined above). Computational modeling looks at markets as a complex system, which is a true function of the market itself.

One final note: Linnea is now a staunch Packers fan and Jack got an iPhone for Christmas.

References

Carley, K. (1999). “On the evolution of social and organizational networks.” In S.B. Andrews and D. Knoke (eds.) Vol. 16 special issue of Research in the Sociology of Organizations. On “Networks In and Around Organizations,” p. 3-30. JAI Press, Inc. Stamford, Conn.

Friedkin, N.E. & Johnsen, E.C. (2011). Social Influence Network Theory: A Sociological Investigation Of Small Group Dynamics. Cambridge University Press.

Mahajan, Vijay, Muller, E. & Bass, F.M. (1990). New Product Diffusion Models in Marketing. A Review and Directions for Research.” Journal of Marketing, 54 (1), p. 1-26.

Mahajan, Vijay, Muller, E. & Wind, Y.W. (2000). New-Product Diffusion Models. New York: Springer Science + Business Media, Inc.

Stigler, G.J. & Becker, G.S. (1977). “De Gustibus non est Disputandum.” The American Economic Review, 67 (2),pp. 76-90.