Everyone in the business could name a "true fan" and nobody could count them. Defining fans by what the company cares about — customers who pay more and leave less — and measuring how each service's recency, frequency and intensity relate to that, produces a score that ranks every customer, says which services make a fan, and holds up on customers it has never seen.
"True fan" became an operational target: high revenue and low churn, combined into one standardized quantity. For every service — mobile data, IPTV, the streaming app, music, payments, the membership store — recency, frequency and intensity (RFM) describe how each customer uses it.
Raw usage is skewed and non-linear: the tenth login a month means less than the first. Each RFM variable is cut into levels at the points that best separate the target; a genetic algorithm searches cut points when exhaustive search is too slow. Each level then carries the target's average for that level.
Regressing the target on the levelled variables gives a weight per service and per RFM facet — which services make a fan — and a 0–100 score for every customer. Holdout deciles confirm the score separates churn and revenue far better than counting usage, and the weights tell marketing where usage growth pays.