Making CRM Pay Off

It is hardly a secret that new business channels, competitive pressures and escalating client expectations have redefined the rules for customer service. What’s more, businesses are often indistinguishable from their competitors in their product offerings and are forced to use aggressive price-cutting as a means of sustaining market share.

Having acknowledged the power of the customer, organizations have been reacting to new parameters by implementing a variety of customer-centric measures in running their businesses. Customer-centricity, however, is limited in scope by resource availability. It has become clear that the key to long-term profitability is service allocation to customers in proportion to their contributions to the business.

Customer relationship management (CRM) is widely regarded as a means of addressing the above issues through focusing on and enhancing customer relationships and services as a means for retaining market share and growing revenues. Its implementation requires an infrastructure to monitor customer relationships, analytics to harvest the relationship data for business intelligence, and tactical direction to implement the discoveries in accordance with the business’ priorities.


In fact, organizations have already invested heavily in the technology infrastructure to facilitate CRM. They have bought into relationship marketing, implemented appropriate hardware and software systems, collected and stored data, and attempted to turn this data into information that generates revenues. The best of corporate intentions seem to have, however, yielded less than spectacular returns on the investments. According to a recent study by the Meta Group, less than 10 per cent of the companies included in a recent survey could measure any tangible return on their business investment in CRM. Where did they go wrong?

The researchers with the Meta Group identified a number of reasons for the failure of the CRM initiatives. They reported that the projects in their survey appeared fragmented and lacked customer focus. They also stated that most companies underestimated the value of customer information they had gathered and stored. Finally, in many cases, measurement techniques were not put in place to monitor the implementation for returned value. In other words, a vision for customer-centricity and an information technology infrastructure were not sufficient to secure a high return on the investment.

How then, does a business ensure that its CRM vision is translated into a profitable, operational reality?


A broad CRM vision might be: “The Acme Company will lead its industry in the value it provides its customers in products and services”. An effective means of implementing the strategic intent of such a vision is to “deconstruct” the statement into small projects with crisp, measurable objectives. Such an approach delivers three advantages: it ensures that the organizational units implementing the projects do not lose sight of the overall strategic objective; it identifies issues in specific business operations that may interfere with the projects; and it enables the organization to see value from the initiative in the near term through the returns on the individual projects.

The vision statement above can be deconstructed into a list of expectations from a successful CRM program as below.

“The Acme Company will undertake strategic initiatives:

• to achieve cost efficiency;

• to deliver shareholder value;

• to grow market share;

• to be the employer of choice;

• to leverage the company brand;

• to understand customer needs;

• to design products consistent with customer needs; and,

• to deliver service based on customer needs.

Depending on business priorities and resource availability, the organization can focus on achieving one or more of the list above deconstructed from the CRM vision.


Conceptually, CRM is not a novel concept. Successful organizations have always paid heed to their clients’ needs. However, prior to the advent of data mining in the business world, business decisions were constrained by the spotty information returned by isolated queries against customer data. The problem with the traditional querying of customer data along a single dimension (or along multiple dimensions as in OLAP) is that such analysis is passive and inward looking; the analyst crafts a strategy based on some assumptions on the customers, and queries the data to justify the hypotheses. Such an approach is limited to the analyst’s perceptions of the business, and does not necessarily incorporate the customers’ viewpoint into strategic planning.

Data mining, on the other hand, is an “active”, discovery-oriented analysis that harvests the data for information consistent with the business objectives. In this respect, the strategy that emerges post-analysis is customer directed.

The above concepts are illustrated below as three case studies of CRM projects to address the different business needs of three (fictitious) companies.

Relationship Segmentation at The Great Bank of Toronto

The Great Bank of Toronto was losing customer investment to its larger competitors. It identified its business priority as cementing its position as the bank of choice for its most valuable customers. To identify and understand its most valuable customers, it conducted a project to partition its customer population and grade each for its value to the bank.

Analysis then produced seven segments that were each measured for their respective profits contribution to the bank, and their share of the customer population. These measurements were then plotted as in Figure 1. The value of such visualization was that such a chart could be used to decide potential business actions on each of the discovered segments, as shown in Figure 1.

(Insert chart1.eps here)

Figure 1: Each of the seven segments (numbered 0 through 6) discovered in the Relationship Segmentation Project were mapped to a grid identifying their profit contribution to the bank relative to their size. Business actions on the discovered segments were determined by the location of the segments in the graph, and are overlaid as the coloured patches.

For the ongoing project it was decided that segments “0” and “3”, having the highest per customer value to the bank, were the “Sweet Spot” that the bank wanted to retain. These segments were profiled on their behaviour patterns to identify their product and channel preferences. Based on the discoveries, the customers in these segments were extended special offers. For example, the bank had recently launched a special Internet banking service that offered interaction with the bank’s financial advisors. Some of the customers in the Sweet Spot were invited to join the scheme at a special rate. To gauge the success of the programs in the short term, the investments of the invited customers were monitored against those of a control group that was also picked from the Sweet Spot, but not extended the special offer.

Next Likely Click Project for WiredSpokes.Com

WiredSpokes.Com is an Internet retailer of bicycling related products, catering to the North American cycling community. Its business objective in undertaking this project was to increase sales revenues through exploiting cross-selling opportunities to its site’s visitors.

To implement this solution, WiredSpokes.Com had to understand the needs of its customers. In this case, the retailer had a loyal set of registered customers on whom it had maintained detailed access logs. These data were harvested to categorize these customers according to their purchasing powers and to identify associations among their product purchases. The result of this exercise yielded a matrix similar to the one shown in Figure 2.

(Insert chart2.eps here)

Figure 2:’s products were binned according to their purchase association with other product categories offered by the retailer and by the purchasing power of the customers who had purchased them historically. The matrix above is a visualization of the binning scheme.

Each bin in the matrix listed products that had been historically associated with purchases in the category indicated on the horizontal axis, for people with the purchasing power indicated on the vertical axis. The cross-selling campaign was based on this matrix.

The cross-selling implementation is exemplified in the flowchart shown in Figure 3.

(Insert chart3.eps here)

Figure 3: The above chart plots a sample path taken by a customer drilling through the Web catalogue to his/her selection, a bicycle tire. The “cantilever brakes” and “heart rate monitor” bubbles are cross-sell opportunities associated with this item that are co-presented to the customer at the end of the path.

In this example, the customer was interested in a specific tire. The drill-down took the customer through “Road Bikes” as the bicycle category, “Tires” as the product category, “650x22mm” as the tire size, and “Specialized” as the manufacturer. The associations matrix was used to identify two cross-selling opportunities – “Cantilever Brakes” and “Heart Rate Monitor” in this case – which were presented as recommendations.

Success of this project was gauged by the company’s revenue growth over the quarter following the solution implementation, and by the percentage of sales attributed to cross-sells over this period.

In the next phase of its CRM journey, WiredSpokes.Com started marketing the cross-sell opportunities to its vendors as “prime shelf locations”, and further trimmed its costs for the goods sold.

MicroMegaTech Applies CRM Internally

MicroMegaTech Computers put a spin on customer relationship management by using CRM analytical techniques to deal with a problem of employee turnover. In spite of offering salaries well above the market average and generous employment benefits, the company was losing employees to its competition. The attrition impacted the firm as lost intellectual capital as well as the costs incurred by new employee recruiting and training. Ultimately, MicroMegaTech’s business objective was to be the employer of choice in the hi-tech industry in its geographical locale. In an effort to identify the source(s) of employee dissatisfaction, it carried out a survey of its employees. The survey gathered information on the employees’ perception of MicroMegaTech, their satisfaction levels on various criteria, and their profile with respect to the company framework.

The survey questions could be categorized as subjective evaluations and as fact statements, but could each be answered by responding on a scale ranging from 1 (High) to 5 (Low). Figure 4 is a sample of such questions in one of the surveyed categories – “Education and Training”.

(Insert chart4.eps here)

Figure 4: A sample of the survey questionnaire filled in by MicroMegaTech’s employees. Question 1 is an example of a question that requires a factual statement in the response, while Question 2 requires a subjective evaluation.

The survey questions that required subjective evaluations could be binned in one of three broad categories – assessment of working conditions, assessment of management, and assessment of the potential for professional growth. The survey data were segmented on the satisfaction ratings in these three categories. The segments were then profiled to identify elements that were peculiar to the dissatisfied segments.

The success of this project was measured as the percentage of the people in the dissatisfied segments that improved their satisfaction level six months after initiation of business actions to implement the discoveries of the segmentation.


For any CRM project, the perspective must be that the long-term objectives of the company centre on revenues and growth rates, and CRM exists only because it is an effective means for growing profits. The CRM projects implemented by the business need to be monitored to ensure value is returned to the organization. In short, the CRM journey should be self-propelled, with each tier of implementation injecting the revenues to take the business to the next level.

Varun Madhok is a consultant with Business Intelligence Solutions, IBM Canada Ltd. He specializes in the implementation of analytical CRM solutions, and in data quality assurance. Mr. Madhok can be reached at