There is a story that runs around about a Friends and Family plan that BT (formerly British Telecom) was promoting in 1999. The story tells that BT designed a discount rate plan for the most frequently called destinations. BT notified prospective customers of this program by sending them their most frequently called numbers. One woman received the letter and uncovered her husband’s cheating. She threw him out of the house and sued for divorce. The husband threatened to sue BT for violating his privacy. I was never sure if that story, frequently used to discuss privacy and ethics in data mining, is a mere urban legend but it always added few smiles to the discussion.
Funny enough, nowadays when everyone wants to harness web2.0 to the benefit of its business this BT story becomes relevant again, at least in the Telco arena. Many communications service providers unofficially try today to augment their customer relationship by analyzing hidden relations in calls data records. The simple example is to identify an “intimate” group of subscribers that communicate frequently and a lot and then assume that there is some strong relationship and influence among them (e.g. family members or close friends).
The next level of analysis is to identify in such groups decision makers or opinion leaders. Telcos have many additional data to guess those things, from location and lifestyle to detailed customer information. The size of the discovered groups and the “social importance” of individuals (e.g. parents vs. the kids) can be then used to augment key performance indicators calculation such as customer value. For example, if we have two similar customers that are likely to churn, a “social network” approach will prioritize a customer that seems an opinion leader in a large group that contributes large revenues. The assumption is that loosing such customer will drag also the loss of his close circle so the loss goes beyond the “seen loss”.
In a similar manner this can be used to identify lost decision makers that should be targeted for win-back or for deciding how much to spend on keeping them loyal. Using social network analysis (SNA) for churn is prone to have even more value in matured and saturated markets, where everyone has a phone and losing a customer probably hides the loss of few other “influenced” subscribers.
Going back to the BT story, for sure taking such analytical approach does not confer a positive image to a service provider vis-à-vis his customers. The legal aspects of it probably differ among countries but anyway I do not know of any service provider admitting doing that. Publications in this domain will probably stay in the research arena until the next wife or husband sues someone.