Companies that have taken to super crunching have begun generating their own data by running randomized experiments for new products and marketing approaches. They randomly split prospects into two groups and check which approach works best, then use regression to find the relationships.
Why do this? Causation! By looking at historical data, it is difficult to determine causation (many variables). By looking at a large random study, you don’t need to control for causation if the groups are equally representative (and you have a sufficient sample size). When you see one group reacting differently, you know the cause is the different approach taken with that group.
Conclusion: Use the randomized data approach to bring your customer’s voice into marketing decision making; remove the gut instinct or intuition of the marketing department and consultants. By listening to the consumers themselves, you will make the most profitable decisions.