All The Way To Bright

February 8th, 2012 by: DSA

Some time ago, I came across the expression “All The Way To Bright”. I think it came out of Citibank, but I’m not sure. Anyway, it stayed with me because it meant, I think, what if we got it all right. What if we’ve finally developed the tools and procedures necessary to manage and grow a direct marketing
business?
 
I think we’re getting close to being there now, at least with regard to contact strategy, or if you prefer, Customer Relationship Management. Here’s what we can and can not do. 

We can create customer segments using behavior and demographic data to create the segments, based on samples, and we can assign all of our customers to these same segments using some combination of regression, CHAID and discriminate analysis.

We’ve learned that creating segments based on attitudinal data is a problem when it comes to assignment (it’s very hard, the errors are too large), but we’ve learned that we can research our behavioral/demographic segments for attitudinal differences and if there are differences use the findings to modify our offers and creative presentations.

We can share and integrate customer data, gathered at various points, so that anyone that deals with the customer can have a fairly accurate picture of the customer’s total relationship with organization. (The more you’re willing to spend, the more current this data can be. Whether direct marketers require access to real-time data is an open to question – if you’re selling airline seats the answer is obviously yes, but if you’re selling insurance, customer service may require real time, or very close to It, data, but marketing is another question.) 

We can respond to incoming customer calls because we can estimate a customer’s incremental lifetime value, and respond accordingly should that customer choose to contact us for any reason.

We can use product affinity models to suggest which customers should receive which offers through direct mail or telemarketing, or when they come back to our web sites. (A word of warning here. One of the issues in a past telecommunications company strike, had to do with customer service representatives that were frustrated because they had to make obviously incorrect recommendations based on a computer analysis of the customer’s profile. Marketers need to take a very hard look at the product affinity models before signing off on their use.) 

We can, in some cases, create customer loyalty programs and/or customer reward programs that improve retention and repeat-purchase and more than cover their costs. (But we’ve also learned that reward programs can’t make up for an uncompetitive price, or unsatisfactory service.) 

We can update our customer databases with information from our Web sites, just as we update our marketing databases with information from our physical fulfillment systems. (But we really
haven’t learned what information is truly valuable and what information is just interesting
.)

We can manage our databases and campaign management procedures to handle an unlimited number of business rules regarding contact strategy. (For example, don’t call the same person more than once in a month for any product; don’t call the same person more than twice in three months for the same product; don’t mail the same offer to the same person more than x times in 12 months, and on and on.) 

We can predict response and lifetime value by combinations of product, offer and channel and thus we are able to predict, with some degree of accuracy, expected profits per name contacted.

We can optimize our selection process so that we can maximize response or sales or contribution within some defined promotion period, given a complex set of restraints and conditions. (For example, Minimum and maximum requirements for any product or product/channel combination; overall min and max constraints for any time period.) 

We can automate, for lack of a better word, all our testing strategies, so that once our databases are updated, our marketing plans can be executed in a fraction of the time it took to implement promotions only a few years ago.

We can tract the effectiveness of our strategies and evaluate which strategies work, and which don’t. (This is the never-ending task of the direct marketer – to find a better way – through
analysis testing of alternatives.
 

We can spend more time on data analysis and strategy development and less time on data manipulation and routine data processing that often pass for marketing analysis.

And, all of the above should make us much smarter marketers. 


Leave a Reply

You must be logged in to post a comment.


Rss Feed Tweeter button Facebook button Technorati button Reddit button Myspace button Linkedin button Webonews button Delicious button Digg button Flickr button Stumbleupon button Newsvine button Youtube button