Myabe an obvious one, but the nature of your business and *who* your customers are plays a big part in what a feasible 'customer value' might be.
I work for a Telco, and conduct a lot of churn/attrition analysis. Also cross-sell models. As part of our data mining for we use customer spend over the prior three months as an indication of customer value. We also look at the number of minutes and number of calls in conjunction with customer spend get an idea of profitability of that customer.
One thing I was pondering for future projects is the use of clustering or segmentation to compute average value for specific groups of customers. Identifying customers those customer value might be lower than average compared to other customers of similar nature (it might be easier to get these customers to increase spend or profitability).
Another value of looking at a customers is not only by the revenue he gets you but rather what is his actual cost v/s his billing it is a sort of profit margin anlysis on each customers. The same can be carried out across different revenue buckets to find low contribution customers and target them for specific schemes & offers