Social network analysis (SNA) is helping companies map and understand the links, associations and possibly behaviors of customers and employees. In the following hypothetical situation, we'll explore the ramifications of using social network analysis in marketing processes and attempt to discern if SNA is "hype" or a valuable tool.


Suppose you are a marketer at a wireless telecommunications company in Europe. Your company has a deep historical record of customer transactions, products purchased and billing history. Three months ago, your company also finished an implementation of an activity based costing project complemented with a profitability management application that now shows customer spend, margins and life-time value.

Deciding it's high time to determine which of your customers are creating and potentially destroying value, you start analyzing a key segment of customers.

On one particular day, while finishing the last sip of your Monster energy drink, you've zeroed in on customer "Thomas Smith".
The analysis shows that Mr. Smith is chronically late on his bills. He does pay but often late, and while you appreciate the incremental revenues associated with late charges, you also examine that he constantly uses your call center to ask mundane questions (bypassing the automated systems). You also see from your product analysis that he likes to switch phones frequently, bringing back product just before the thirty day exchange policy expires.

In the initial analysis, it appears Mr. Smith is costing your company a lot of money. Now, as a marketer, should you keep Mr. Smith as a customer, or gently pass him onto the competition?

What might be obvious is in fact a very complex decision, especially when you include social network analysis.

Leading edge companies are using Social Network Analysis to detect and interpret the patterns of social ties within a customer base. Authors Stanley Wasserman and Katherine Faust in their book "Social Network Analysis: Methods and Applications", mention that SNA is "based on an assumption of the importance of relationships among interacting units. The social network encompasses theories, models, and applications that are expressed in terms of relational concepts or processes."

In this particular marketing example, social network analysis can be used to determine the "importance" of Mr. Smith, especially in relation to other paying customers.

So let's get back to the analysis on Thomas Smith. By adding call detail records to your data warehouse, and using social network analysis techniques, you now see that Mr. Smith is a "node" in a pretty complex network of customers. In fact, through a "Fab Five" campaign you concocted a year ago (where customers can call five friends for free in-network), you also see that Thomas is linked to five very profitable customers.

Since Mr. Smith is well connected to five very profitable customers, treating him poorly on his next customer visit, or jettisoning Mr. Smith altogether could lead to the defection of his five closest friends ... customers on your network, and customers that pay their bills and produce positive cash flow.

Let us suppose you also had the divine prescience to add net promoter scoring to the mix. Now you notice that while Mr. Smith has trouble paying his bills, he is in fact a "promoter" of your company. He likes your friendly customer service representatives, and also is very generous in telling his friends about your willingness to "bend over backwards for him" to meet his changing needs.

With this analysis in hand, you determine that Mr. Smith isn't a customer to jettison, and in fact, is probably one that deserves a closer look.

Before we get too caught up in the hype, social network analysis isn't a savior to marketing decision making.

Social network analysis–done right–requires a lot of data from myriad sources. In the hypothetical marketing example above, simple call detail records are used, but to say–find a terrorist–the National Security Agency of the United States (NSA) would require call detail records, credit card transactions, car rental receipts, and many other digital markers. Even then, the output of the analysis isn't always accurate.

Accurate data is a key factor in reliable results, but so too are the assumptions used in the model. Is Mr. Smith really that "important"? Should we assume that "associations" are actual close relationships? Should we also assume that if Mr. Smith leaves our company, his "Fab Five" connections will follow?

Social network analysis–hype or help? It's up to you to decide.

Questions:
* It is well documented that Social Network Analysis (SNA) can be the source of "false positives". Would you trust the output of SNA to make marketing decisions in your organization?
* Call detail records show Mr. Smith calls five contacts quite often. Does "activity" ... in this instance a phone call–denote a "close relationship"?
* SNA can also be used to model employee connections in an organization–in effect to determine the "importance" of an employee? Good or bad idea?
* If you were the marketer in the situation above, what other tools might you use to determine if you should keep Mr. Smith as a customer?