Wharton’s Gad Allon discusses his research on using social media data to identify a company’s top customers.
Listen to the podcast:
Social media offers an almost endless stream of data for businesses to collect on their customers. But what good is data without a smart way to apply it? The latest research from Gad Allon, Wharton professor of operations, information and decisions, offers a lifeline for firms drowning in the deep waters of social networks. Allon and his team devised an analytics model that can help businesses identify high-value customers. The paper, “Managing Service Systems in the Presence of Social Networks,” was co-authored with Washington University professor Dennis J. Zhang. He talked with Knowledge@Wharton about the process.
An edited transcript of the conversation follows.
Knowledge@Wharton: Tell us about what you studied and the main questions you examined?
Gad Allon: We’re looking at the intersection between social networks and service providers, and we’re asking, “How should service providers think about these social networks?” Now, social networks have existed for many years. Think about reading clubs, churches and so on. But the emergence of the new social networks, such as Facebook, Twitter, Instagram and Snapchat, allowed firms to not only look at what customers do, but track it over time.
Nowadays, firms have presence on these social networks. But more importantly, consumers use these social networks to communicate among each other and tell others about the experiences they had. Why is that interesting? In the past, we had key opinion leaders who had access to maybe The New York Times or The Wall Street Journal and wrote there about food and so on. Now, we have more and more people that have access to, let’s say, 1,000 followers on Twitter. So, we have to think about the idea of influence in a more nuanced way.
“Technology enabled firms to micro-target and offer very segmented, differentiated services to customers.”
That’s on the social network side. From a service provider side, technology enabled firms to micro-target and offer very segmented, differentiated services to customers. They can offer upgrades based on who you are, your history, what you’ve done. The combination of the two — the fact that firms can look at customers not only in terms of the value they bring, but also how they influence others on social networks — poses very interesting questions. The main question we’re trying to answer is: How should service providers allocate scarce resources, whether these are priorities or better service, to customers based on their value, but also their social network presence, and how they’re influenced and being influenced by others?
Knowledge@Wharton: What did you find, and how can your findings be applied in business?
Allon: First, we built a model of how customers tell others about their services, how they form beliefs about the quality of the service based on their own experience, but also what other people tell them. We’ve identified and indexed an idea we call “centrality of the customer.” The economically adjusted centrality of the customer. It looks at not only the value they bring, but also how they influence others and how they are being influenced by others.
The main takeaway is that firms should think about ranking their customers according to this index. Essentially, a customer should get high priority, or a better service, if they fit into one of three categories: they have many friends; they have very few friends, but these friends bring very high economic value; or this customer himself brings very high economic value, but is not being influenced by others. If he’s being influenced by others, I’m better off giving it to one of their friends.
One of the main things that this doesn’t have in it is this idea of “friends of friends.” As you notice, you need to look only at the closed circle. That stands in stark contrast to what we know about products or political opinions. The reason for that comes from the nature of services. The nature of service means that I’m offering a customer a service only when they are there, which means that I can provide a very differentiated quality of service, which means that when the customer comes, I don’t have to really think about how this information is going to propagate. The moment they bring their friend, I can then decide on the right quality of service to provide. What it means is that it makes it much easier for a firm to track and to combine this information into the decision-making process.
Knowledge@Wharton: How does this interface with the concept of customer lifetime value? It sounds similar.
“While many firms track the lifetime value of their customers, very few understand how their customers interact on social networks.”
Allon: While many firms track the lifetime value of their customers, very few firms understand how their customers interact among themselves on social networks. Should firms really go and obtain this information? We say that for firms to obtain this information, this information needs to be valuable. We look at the correlation between the economic value and social value. Surprisingly, we see that for some businesses, there is positive correlation. This means the more vocal customers are also the ones that have higher lifetime value. In which case, it’s pretty easy. I know that I need to target those with high lifetime value.
However, the situation we see in many businesses is that it’s negatively correlated, meaning you have highly vocal customers and customers with very high lifetime value. In this case, knowing the social network and how it interacts with the lifetime value is crucial. In fact, we’ve obtained data from Yelp that showed for restaurants that are above-average in terms of their pricing, they’re very likely to have negative correlation between the two. In which case, knowing not only the lifetime value, but also the exact way that your customers interact with their friends, is going to be crucial to obtaining better value from providing better service to the right customers.
Knowledge@Wharton: What’s next for this research?
Allon: The main question we try to answer now is: How should firms think about the difference between pure information and opinion on these social networks? There are many more markets for influencers, what we call key opinion leaders, where firms try to ask these people to tell their friends, tell other people, maybe celebrities, tell their audience about a product. But what part of that is purely informational? The goal is to bring additional volume. What part of that is really trying to shape customers’ opinion?
Clearly, if you hear about a celebrity that got an amazing service from United, you’ve heard of United before. The question is, can this information shape your opinion about a firm that you already know something about? Is it only for an emerging market, or is it for a market that already exists? What we try to do is, both empirically and theoretically, disentangle these two effects.