As with other changes that are occurring in our society today, technological advancements have created the environment in which more players can play the game. In the late seventies or early eighties, only the large direct marketing businesses such as Reader's Digest actually practiced CRM and data mining. They were able to conduct these practices by investing millions of dollars into the development of an appropriate marketing database. The investment into this technology was premised on the notion that Reader's Digest would be second to none in terms of understanding its customers. Certainly, the investment in this technology was justifiable since direct marketing was its core business.

Direct marketing often gets a bad rap as being simply "junk mail." But the principles of successful direct marketing are really what CRM and data mining are all about. As practitioners of successful direct marketing, the Digest always knew who to promote, what to promote and when to promote. However, for most other organizations, this was not the case as direct marketing represented only a minor portion of a typical marketing budget. The business paradigm for most marketers during the late seventies and early eighties was optimization of revenues that was supported by the prevailing business culture for that time period.

Yet, the technological advancements in the last twenty years have not only changed how society functions but it has also ultimately changed its expectations. With technology, we are expected to do more with fewer resources. Consequently, most of us are constantly time stretched as demonstrated by the growth in books and seminars on time management.

From a marketing perspective, this has been manifested by the decreased time that people have to focus on a given topic or communication. At the same time, technological advances have facilitated the entry of organizations into direct marketing. It has become less cost prohibitive to invest in the required resources for successful CRM and data mining. Marketers have also taken these principles used so often within the direct mail channel and are now bombarded with all kinds of communications and offers in the hope that consumers will respond appropriately to each company's message.

With the presence of the Web becoming an ever-increasing marketing channel, we now have the ultimate medium whereby the collection and actioning of individual information can occur dynamically. This means that decisions regarding how we communicate to customers in this medium can change instantaneously. In the old direct marketing environment, we had to wait for several weeks after the launch of a campaign before we actually received any information from consumers. The dynamic and interactive nature of the Web will require that CRM and data mining become more firmly entrenched as ongoing business practices.

The increased access to information and more importantly the capability to use it is a significant competitive advantage in today's business environment. Organizations operating under this knowledge-type environment can view all their marketing initiatives from a ROI standpoint. At the same time, these organizations can gain tremendous insights into what worked and what didn't. This dynamic learning process provides an environment whereby the business is always striving for improvement. This is the key to successful data mining and CRM. The cycle of campaign execution and learning provides the ongoing feedback to constantly improve future marketing efforts. The best way to illustrate this, though, is through a live case study whereby we see how an organization has evolved through its CRM/data mining practices.

In the eighties, American Express was an organization striving to increase its membership base. In fact, its share price was directly impacted by how many new cards were acquired. Towards the end of the eighties, CIF (Cards in Force) targets were being achieved but at a rate where the cost per new card doubled. These cost inefficiencies were unacceptable to the organization. In order to resolve this dilemma, they began to purchase names from lists that seemed to fit the profile of an American Express cardholder. Lists such as Business Week, Forbes, Fortune, etc. produced very good acquisition results. However, the number of names on these lists were simply too small to provide the required prospect universe that was necessary to generate Amex's aggressive CIF targets. Ultimately, Amex needed a very large prospect universe but using tools which allowed them to generate new cards in a cost-efficient manner. The solution was twofold.

First, they built a prospect history database that compiled information at the individual prospect level. Examples of some key information included the promotion history, and gender of the prospect. With this database, Amex began to bucket its prospects into two key segments based on promotion history. The two key segments were new names and previous names.

The second solution was to then build predictive models that could be applied to both universes in order to optimize response rates. Using overlay aggregate level data from Statistics Canada, gross response models were built and provided a 50% lift in acquisition performance. However, learning revealed that approval rates had actually deteriorated. They had indeed discovered that the best responders were those most likely seeking credit. The learning from this campaign indicated that Amex needed a targeting tool that optimized net response rate (gross response and approval) and not just gross response. A net response model was developed and once again, they achieved a 50% lift in acquisition performance. However, they achieved much better cost efficiencies within their operations area, which dealt with new applications.

Yet, the learning also revealed that, though, they may indeed be optimizing net response rate, these new card members may not be profitable. In other words, they may not be using the card or worse simply cancelling it. As a result, Amex decided that the goal of acquiring new cards should be based both on their net responsiveness as well as their overall profitability within their first year. By looking at prospects in this manner, they were essentially able to evaluate each prospect based on their predicted ROI (Predicted Profitability/Predicted Net Response).

This above example demonstrates how an organization evolved their overall acquisition strategy based on data mining. The decision-making process in each stage was based on results and also provided input on what the organization needed to do in order to achieve the next level of improvement.

It is important to understand that the same learning feedback process used to generate strategy improvements can also be used in a more granular fashion by identifying specific tasks and tactics, which can be improved for future marketing efforts. Besides the notion of ongoing learning, successful CRM and data mining practices include the application of this learning for future activities. The old adage of "Analysis Paralysis" is a trap that marketers can easily fall into whereby nothing is ever actioned. The objective of any marketing environment is an ongoing learning cycle whereby the testing and tracking of programs results create new insights, which are then actioned in new programs.

Richard Boire is a Partner of Boire Filler Group, a Data Mining and Customer Relationship Management company.