CRM: Picking Up the Crums

By ManMohan S. Sodhi

Much has been written about the "crumminess" of customer relationship management (CRM) solutions. Thirty percent of CTOs in InfoWorld's recent CTO network survey said that CRM was the biggest blunder/most hyped technology of 2001[1]. CIO Magazine had a tell-all article [2] and the Economist referred to CRM as having "failed to live up to its promise" [3].

Parallels with SCM

Of course, there are many reasons — vendor-related (buggy, non-existent or mislabeled software), implementer-related (poor project management or poor implementation) and customer-related (unclear goals, unrealistic expectations) — but it is worthwhile taking the higher ground to see the technology evolution forest, not the software trees. I find interesting parallels between supply chain management (SCM) and customer relationship management (CRM) evolution in three phases. For SCM, these were:

Phase 1: Point solutions for different domains, including those in supply chain planning running operations research algorithms.

Phase 2: Integration of supply chain information using enterprise resource planning (ERP) software from SAP and others.

Phase 3: Ways to extract value out of the integrated information, in particular, solutions from i2, Manugistics and SAP among others, to improve planning.

Likewise, for CRM, there were and still are point solutions like database marketing and data mining techniques. Then came the integration phase for providing one face of the customer to companies by integrating all customer-specific data in one place with solutions from vendors such as Siebel. This is also the phase with failed implementations and low or non-existent return on investment. This was the case with ERP as well, and the debates of whether or not ERP implementations provide or provided a positive return on investment still continues. The third phase, already in swing, is in the use of advanced algorithms for extracting value out of the integrated information. And this is where operations research steps in. Rather than discuss algorithms (see [4] for a start), I will outline an overall context for CRM.

With most customer data in one place using software such as Siebel's, the marketing team in a company can use the data to design new offerings for reconstructed market segments in order to garner more revenues through additional sales and/or through higher prices. The third-phase CRM software from companies like Data Distiller, Norkom, Quadstone and E.piphany goes further in providing the customer service person with real-time recommendations for the individual customer regarding the propensity to buy another product or service. This is similar to the up-selling or cross-selling a customer service representative for a catalog-based clothing retailer may do.

CRM Objectives

CRM should either decrease costs or help bring in new revenues. We can increase or protect existing revenues using CRM by:

- designing better, new offerings tailored to improved market segments;
- preventing customer attrition through improved service, offerings and fulfillment;
- generating higher average price through revenue management;
- preventing excess volume discounts; and
- up-selling and cross-selling.

Potential buyers of CRM software may be able to achieve the above with better use of software they already have. This will avoid risks associated with technology, skills and people when acquiring new enterprise software. So it is good to understand objectives.

Modeling and the Overall Context

Both potential buyers and would-be vendors of CRM software need to think about such issues as:

Granularity and associations for stored information: Besides the level of detail and how to tie the data together, this also means defining the "customer." This may be harder for industrial customers than it is for end consumers.

Dimensions of analysis: There are many dimensions of analysis, as the software users may want analysis at the customer level, channel level or geographical-unit level in their search for segments.

Building and testing hypotheses: This includes questions as which hypothesis to study, and what comprises the control group.

Type of customer: The customer-type is crucial as different types have different interactions. (See [5].)

Pricing and revenue management: While the airline industry has been sophisticated about pricing, other industries are trying to catch up.

Real-time versus static use: Static use includes improved market-segmentation. On the other hand, analysis can produce intermediate results, which can be used for algorithms that produce results in real-time.


- InfoWorld CTO Network Year-in-Review Survey, Dec. 1, 2001 ( hn/xml/01/12/24/011224hnyearend.xml).
- Patton, S., "The truth about CRM," CIO Magazine, May 1, 2001 ( 050101/truth_content.html).
- "Analyze This," March 8, 2001, The Economist.
- Sodhi, M., "Sensing and responding to new marketing and customer service opportunities," Cyberspace, OR/MS Today, August 2000.
- Sodhi, M., "Partner relationship management," Cyberspace, OR/MS Today, June 2001.

ManMohan S. Sodhi ( is vice president at Gandiva.