Using Automation To Create Corporate "Memory" of Your Customer Relationships

We have seen the development of an awful lot of technology which appears to be oriented towards simply remembering what we've said to our customers and what they've said to us.

We know that recognition and remembering are important foundations for relationships, but the volume of data is very large in comparison to the volume of pertinent information. For example, weblogs are interesting for seeing navigation patterns which may point to site design opportunities, but in most cases the data is not terribly interesting or important to the relationship.

Most analytics also work with poorly defined events that they are supposed to predict, and have poorly defind events as predictors as well. For example in Banking, most can't tell the difference between new money in an account and money that switched in from a different product held by the same customer.

One of the things we are doing is taking a step back from the data and processing it into information before analytic techniques are applied. The value of this, we have discovered, is large (as much as 30% improvement in targeting, for example).

For discussion:

Are you pre-processing data into information before applying statistical and other analytic methods ? Why or why not ?

*****UPDATE********

Given the null response to this question I suppose the answer is self evident. I am somewhat surprised that no-one else is working on this, given the uplift we have seen from doing it. Green field... that's fine with me !