Facebook’s Trials and Tribulations
Two recent articles about Facebook highlight the importance of understanding the consumer better in order to facilitate a match with advertisers on the company’s platform. The first, more innocuous article “Facebook’s New Ad Effort Focuses on Emerging Markets” in WSJ discusses the importance of understanding differences in behavior across countries to better align marketing messages to target customers, citing as an example the frequent practice in India of giving friends a “missed call” (calling them and hanging up after a pre-arranged number of rings) in order to communicate a specific message (e.g., “I have reached home”) without having to pay. How can such behavior be translated into action? Facebook allows consumers interested in a certain brand, say L’Oreal, to give a phone number associated with that brand a “missed call.” In exchange, consumers receive a call from the brand with a targeted message, video or promotion. This way, the advertising campaign piggybacks on a routine behavior of the consumer to accomplish a specific brand goal, with Facebook acting as the link between the two. Further, since the basis of the interaction is the customer’s choice to receive the message, the platform enables L’Oreal to better target its potential customer.
The second, more incendiary WSJ article “Facebook Experiments Had Few Limits”, was based on the revelation that Facebook’s data science group ran an experiment in 2012 in which the news feeds of nearly 700,000 Facebook users were manipulated to show more positive or negative posts. The study found that users who saw more positive content were more likely to write positive posts, and vice versa. Although Facebook may have been legally in the clear when running such an experiment, changing how users feel is certainly not justified on moral grounds. I will not debate the merits of running such experiments, although as I have noted in the past, the methodology of running experiments is often central to understanding the causal effects of marketing interventions. However, my focus here is why Facebook might be interested in the outcomes of such experiments. To understand this, let us compare what Facebook was seeking to accomplish in these two articles.
“Facebook’s New Ad Effort Focuses on Emerging Markets” has Marketing 101 written all over it. Specifically, it is based on the notion that there are cross-sectional differences across consumers, and as we gain a finer understanding of these differences, we are better able to deliver customer-specific value propositions. After all, this is, what the concept of segmentation – targeting – positioning is all about, albeit at a very “micro” level - and marketers are constantly getting better at this. To paraphrase Optimus Prime (Peter Cullen in the Transformers movies): “There are innumerable mysteries to the universe. But who we are, is not one of them. That answer lies inside us.” As marketers get better at targeting however, the question becomes: how do we go beyond cross-sectional differences across consumers to gain a deeper understanding of each consumer?
Based on the recognition that the same consumer can behave differently at different points in time, “Facebook Experiments Had Few Limits” tackles this question. A given consumer might behave one way at time index t compared to how (s)he would behave at time index t+1. How a given consumer behaves at different points in time then becomes the basis for this temporal segmentation, i.e., the unit of analysis becomes the consumer-moment rather than the “cruder” consumer level.
Marketers have definitely recognized this time variation in consumer behavior. In particular, there is considerable research indicating that the context in which I consume determines my choices. Research I did with former student Minki Kim (now a professor at KAIST in Korea) finds that brand preferences of consumers across various social groups and context-related scenarios could differ dramatically. For example, when drinking beer alone, I might drink Budweiser, whereas with friends I might prefer Corona. Similarly, when I drink beer at a local pub, my brand preferences could be very different than when I am on the dance floor (see the full article in QME). Why is understanding these differences useful for marketers? When a brand has a small share (or when its overall preferences are small), such an analysis identifies specific scenarios within which there might be strong preferences. For example, in scenarios where consumers are working at home with a couple of other people or dining with friends, Corona is the highest-ranked brand although it has only a 4% share in our data. This information can then be used by the brand to reinforce its image, and can provide the basis for advertising campaigns by identifying the kind of copy or imagery that might be used in commercials. This way, firms can use the information to facilitate the positioning function.
One can immediately see the problem with undertaking the above analysis for each category and for all conceivable contexts consumers might find themselves in. This would be a herculean task, even for Facebook. However, if one can understand why I prefer a brand of beer in a given scenario, I can then see whether this reason carries over to my preferences for say, a brand of pizza in the same or different scenario. This is where the Facebook study comes in - by first understanding how consumers behave when they are positively or negatively disposed, and then relating this disposition to the various contexts or brands the consumer may engage in, Facebook can reduce the previously herculean task to one that is more manageable. For example, if I am in a good mood, I tend to hang out with friends, drink a Heineken, and consider a vacation to the Bahamas. Knowing this, Facebook can now look for signs of such a disposition in the future and suitably target appropriate marketing messages to me. In other words, Facebook is able to tackle the consumer–time index segmentation problem.
In Divergent, when Beatrice ‘Tris’ Prior (played by Shailene Woodey) gazes in admiration at Four’s (Theo James) tattoo and says, “It’s amazing. The factions. Why do you have all of them?” Four replies “I don’t want to be just one thing. I can’t be. I want to be brave, and I want to be selfless, intelligent, and honest and kind. Well, I’m still working on kind.”
The consumer is complex and far more than a monolithic creature. If marketers are able to decipher this complexity, they would be able to do a better job delivering value.