SNA’s Failures Revealed – Getting it Right With Social Ripple Analysis

If you are using one of the market leader Social Network Analysis (SNA) tools, it is very likely that you are not seeing the forest for the trees. While SNA is an effective tool for identifying subscribers who are social leaders and followers, using it for churn prediction and propensity modeling can be dead wrong. As Forte, we recommend all operators take a leap towards using Social Ripple Analysis (SRA)…


Social Network Analysis (SNA) is undoubtedly one of the rising stars of customer analytics (primarily in telecoms), bringing in the impact of social relations on customer decision-making into the picture, as well as improving various existing models (such as churn prediction and cross-sales modeling). While the benefits realized by most users are unquestionable, its traditional applications oversimplify the social paradigm and can guide companies towards making misguided / incorrect decisions.

Companies that have been using SNA should now take the next steps towards improving their understanding and accuracy in modeling of social interactions, by focusing on “social diffusion” models. Social Ripple Analysis (SRA) improves (moreover, complements) SNA solutions by leveraging this perspective.


It is a given that consumers listen more to each other than they listen to companies when it comes to making purchasing / company relationship decisions. According to a survey in 2007, the top source for product information was word-of-mouth, with a whopping 59% of customers referring to each other as the key method for making decisions. While SNA has jumped in with both legs to save the day, there are certain factors many SNA models do not address properly (despite a commonly held belief that they in fact do).

Below are three SNA myths that should not be taken 100% at their face values:

SNA Myth #1 – Leaders have “more influence” on members of the community

The TruthLeaders have influence on “more members” of the community

How Come?


The leader of a social community is frequently the one with highest number of connections within the community. Although this means that a leader can affect “more” individuals with his / her decisions, it does not necessarily mean that every member of the community will be influenced equally with these decisions. Often, a single member’s top influencer can be different than the leader itself; in the above illustration, for example, John’s wife could have more influence on him rather than his former classmate James (whose phone number he does not even remember).

Most SNA users would argue against that their predictive models are improved by focusing on the status of community leaders. Although on a high level the math holds, it is not necessarily for the right reasons. Let’s assume, for the sake of above illustration, that a churning customer increases by 10% the risk that every other individual he / she is in contact with will churn as well. For the above example, this would mean that:

  • James – Interacting with 6 different customers, this means that there is a 60% likelihood that there will be a churner in this community.
  • Rachel – Interacting with only 2 different customers, this means that there is a 20% likelihood that there will be a churner in this community.

Although James’ churn will have a more drastic impact on the community, it is not due to the fact that each member will be affected more; rather, it is simply because he is in touch with more people. As a matter of fact, his departure would have zero direct effect on John, whereas Rachel’s would increase John’s churn risk significantly in this case.

Ask yourself…who would you take advice from, and be susceptible to being influenced by – your wife, or, the 18 year old cousin of your colleague whom you’ve never seen in your life but apparently is quite chatty with your colleague and their family?

SNA Myth #2 – How frequently you communicate with an individual dictates the level of influence you have over that individual

The Truth: How much you communicate with an individual means how much you communicate. Period.

How Come?


Every relationship between customers is unique, defined by the nature of the relationship, the demographics of the parties, as well as their personal needs and experience with the products and services they use. As a consequence, not every relationship has the potential to create the same influence. Although the strength of communications between two individuals is an important parameter of how much they can influence each other, it is very misleading when used as the sole indicator. Furthermore, the parameters which affect whether a customer would be influenced from another depends on the area of influence (i.e. churn, data bundle sales, tariff migration) itself. In above example, although John speaks most frequently with his mother, when it comes to product usage, he is more similar to his best friend, and is more influenced by his comments. Regarding business related VAS, however, he looks up to his boss, who is a more trendy mobile user.

Ask yourself…who could convince you that the new iPhone deal is a good one – your grandfather, who still uses his Nokia 2100 but you talk to everyday, or, your tech-savvy colleague, whom you don’t call so often since he sits in your next cubicle, but rely on greatly for advice around technology?

SNA Myth #3 – Individuals are influenced primarily by the core “social network” they belong to – each network contains numerous people who are sometimes strangers to each other.

The TruthIndividuals are influenced primarily by their “personal network” – each person has his or her own personal network comprised of people he or she is directly connected with.

How Come?


Although the social networks that subscribers belong to can have a certain influence on them, moreover it is the people that subscribers are directly / closely connected with that have greater influence. A subscriber of a given community comprised of 20 individuals may not have even heard of a product or service that 15 of that community’s members are using (as he or she is not directly connected with those people). On the other hand, just like John in the above example, a subscriber could be easily influenced by subscribers that are a part of his or her “personal network” but outside his or her core “social network.”

Opposition to this myth suggests that assigning customers to distinct communities may not be the best idea, after all. If John has separate relations with friends, family, co-workers, etc., why not put him at the center of his own community, and place his relations around him to get the true picture around who influences him? With such a perspective (which Social Ripple Analysis, the concept we will explain later in this article, takes into account), every individual becomes the center of his / her own personal community.

Ask yourself…would you switch operators because 2 of your and 6 of their not-so-close friends switched to some other operator or because your wife and child just did the same?

It is possible to expand this list further and debunk even more myths; what we have listed, however, should be enough to prove the point that SNA solutions (in and of themselves) are not what they are hyped up to be.

To summarize and simplify, relations are too complicated to be illustrated using simple graphs. The question of how to overcome the limitations of SNA can be answered by Social Ripple Analysis (SRA).


SRA focuses on a true “social diffusion” view around individuals (instead of synthetically creating communities, the way SNA does), and models how much a given subscriber can influence another given subscriber directly around driving up the sales of a given product or service, or around driving the churn of that subscriber.

Deploying SRA involves conducting a six-step effort, which we have turned into our very own solution – Rippler©. If you are an operator and wish to learn about these six steps, or, would be interested in learning about Rippler©, please contact us via

What Next?

As Forte, we are challenging all existing SNA users to hit the pause button and test their current model performances against SRA. Having experienced its impact first hand, we strongly urge companies to close the gaps in their social view of customers using SRA or similar analyses.

Does this mean that the era of SNA is over? Of course not…when used properly, SNA can create significant value for companies. However, companies should start digging below the surface to gain such benefits, which we refer to as SNA 2.0, and includes analyzing and acting at the community level, rather than the individuals themselves. 

Note: For what it’s worth, we still believe that every operator should experiment with SNA solutions. To facilitate this, and since at this point with the development of SRA that SNA is relatively a dead duck, we are now offering our proprietary SNA software for free.

Tags : analytics, business intelligence, churn, consulting, cross-sales, customer analytics, data mining, decision management, marketing, network, retention, sales, SNA, social network analysis, social ripple analysis, SRA, strategy