Unsound implementation strategies often begin with project definition. CRM has been defined at a project that integrates processes across marketing, sales and customer-service divisions, in order to leverage each other's strengths. But this means standardization of business processes, the enemy of innovation. "We saw CRM developing and becoming more like ERP -- very rigid," says Scott. While rigidity may work when dealing with accounting and human resources, "a marketing campaign has to be more flexible and dynamic," he says. "Separating data from processes can make CRM more flexible."

Another early mistake in the implementation stage is to manage CRM centrally and plan to roll it out as a giant, fully integrated solution. This was the way of ERP, which, for the most part, had to operate as a single system before companies could reap real value. With CRM, companies should plan on a series of short, focused projects that are, critically, owned by line-of-business managers. "They're the ones that know the processes," he says.

This, of course, leads to the issue of governance and measurement -- moving targets in the world of CRM. AMR Research reports that 58 percent of CRM projects are launched without a business case or set of metrics to be monitored. But benchmarks are crucial, Scott says. Without them, mediocrity or failure could be viewed as success. Benchmarks also play a key role in creating strong IT governance; IT is a watchman of sorts for the decision-making business managers.

Scott believes that every CRM project should start with an assessment of internal performance metrics and plans for a comparison of those metrics with external benchmarks. The assessment and benchmarking process will identify the root cause of performance problems and help channel investment to the right projects, he says.

The last loaded gun pointing at CRM is vendor selection and integration. Businesses are unique, Scott says, so making the decision to use a single CRM vendor for platform, applications and database to ensure smooth integration, is pure folly. Again, flexibility should be the guiding light.

In short, companies should go with a mishmash of software vendors that best fit their needs. What about integration? Tomorrow's Web services-based architecture breaks down the old way of thinking. The new architecture lets companies separate the data model from the business logic, and the business logic from the presentation layer. All of which frees up data fields and business objects to flow from system to system, rather than being locked into a single enterprise application platform.

Web services though, won't solve the CRM conundrum. In order to turn the tide of CRM failures, "planning, governance, all things have to start with the user," Scott says. On the upside, Scott believes people are finally catching on to this new way of thinking about CRM, "and next year, there will be more successful deployments. Things are getting better, but it'll still be a couple of years before you really see the enterprise-wide, single view of the customer," he says.