Behaviour is another good way to segment customers, and this can include simple things like call frequency, local/international ratios, data usage, etc. It can also be more advanced, and incorporate link analysis, call circles, etc. Identifying customers who are "communication hubs", or early adopters in their personal network can enable very targeted marketing.
Marketing is the lifeblood of a telco, and there are many aspects to this. The most basic building block of an analytical approach to campaigns is the propensity model, estimating the individual likelihood of customer response. Just as important is tracking the success of past campaigns, to gradually improve the methodology, which involves carefully designed control groups.
In addition to propensity, contact rules should be implemented to ensure that the best customers aren't spammed constantly, and the next tier down ignored. A proper campaign management tool makes this much easier to manage. The next step for advanced companies is marketing optimization. This is a way of combining all the propensity scores and contact rules, along with other constraints such as channel limits, budgets, targets, etc, into a mathematical model that optimally assigns customers to campaigns over a planning horizon. It is quite dramatic how much improvement can be made over what seem like smart manual strategies. There are just too many factors involved for humans to pick the best combinations.
Beyond even optimization is real-time. The most advanced telcos are starting to implement models that update instantly with every customer interaction, and can highlight to a customer service representative, or on the internet, an appropriate offer tailored to that customer at that time. This can prevent churn (when desired - some customers cost more than they are worth), and increase up-sell, cross-sell, and customer satisfaction dramtically.