To understand why operational systems generally must be re-engineered to achieve best database marketing practices, consider a hypothetical company that sells to both consumers and businesses. Also assume that:
- Many orders come in over the phone, and B-to-B transactions are much larger than B-to-C.
- The company has invested in an elite group of highly-trained phone reps to handle the lucrative B-to-B transactions.
Now, consider what happens if the following automated prompt is put in place to route incoming calls: “Please press or say ’1′ to speak with a telephone support representative, or press or say ’2′ to speak with someone from our corporate order group.” This unfortunate wording will adversely affect both financial performance and database content because:
- Many B-to-B customers will press or say “1″ before even hearing the end of the prompt.
- Therefore, they will be routed to B-to-C phone representatives who have not been trained in the important B-to-B protocols.
- The operational system will miscode B-to-B customers as B-to-C, and this error will be propagated into the marketing database during the next update cycle.
Techniques exist to correctly reclassify some of these misclassified B-to-B customers, including look-up routines against both keyword tables and third-party compiled databases. However, such techniques are imperfect because of factors such as:
- The use of personal credit cards for company business. (Credit card affinity programs such as airline frequent flyer clubs encourage this phenomenon.
- Calls that originate from SOHO’s (small office/home offices).
- Shipments to a home address, such as holiday gifts that new business development professionals plan to deliver personally while covering their sales territories.
Re-engineering the automated phone prompt will improve the classification of customers, and will enhance marketing database content in the process. Interestingly, changing the prompt to, “Please press or say ’1′ to speak with someone from our corporate order group, or press or say ’1′ to speak with a telephone support representative” is likely to result in an overstatement of B-to-B customers.
Also, care must be taken that improvements to the classification of customers do not cause problems with existing data mining routines such predictive models and customer clusters.
Jim Wheatonis a co-founder of Daystar Wheaton Group.