Weapons of mass marketing

Since the first time a rock was traded for some berries, business transactions and operations have generated data. Century-old accounting methods enabled managers to maximize long- and short-term profits and allowed investors to evaluate alternative investments. Data recorded the movement of goods and labor directly and the movement of money less directly.

Telecommunications service providers always have generated an enormous amount of information during the business processes of customer acquisition, network usage and billing. Historically, that data met operations and back office requirements, not marketing needs. This was a sound business practice because marketing in a monopoly does not require much sophistication.

The incremental profitability derived from incremental computing was harder to achieve when computers were costly. Moreover, once achieved, it often was returned to ratepayers anyway. The techniques that provide useful management information while protecting customers' privacy were not yet available. Now, deregulation and advances in technology have changed the variables of this equation.

Complicating the matter, each year it gets harder to tell what information is part of the operations' function and what is part of customer data management. Switches, programmable or not, are nearly indistinguishable from computers, and computer-based telephony blurs the line between working and networking. As computing becomes more sophisticated, the accounting function of business information still is necessary and active, but it has receded into the infrastructure, while such previously mystical functions as marketing support have blossomed.

New customer data management applications enable telecommunications providers to gain new knowledge of customers in aggregate and apply that knowledge to one-on-one marketing (Figure 1). Facilities-based telecommunications providers are making huge expenditures in support of strategic information systems. These investments conceivably could exceed those for traditional outside plant and network infrastructure. Is this shrewd, frivolous, evolutionary or simply terrifying?

Customers will find the service packages that best meet their needs, and service providers will use technology to apply their marketing resources to those customers that offer the most profit, sending unprofitable customers to competitors to reduce their bottom lines. But how does a carrier decide who is in which category? Customer profiling has provided some answers, and most telecommunications providers are establishing data warehouses to enjoy these benefits on an ongoing basis.

Implementation issues

A data warehouse contains three main elements: customer data, hardware/operating software and analytical tools. Each component presents its own implementation challenges.

Customer data. One might expect that the easy part is assembling the data because much of it already exists in electronic format, but this is one of the most difficult tasks in building the warehouse. For decades, the data supporting disparate corporate applications developed independently. Companies already finished with the implementation of resource planning undoubtedly will have an easier time assembling the appropriate data than those that plan to build a warehouse from legacy systems.

Companies that are the product of large mergers - nearly all by now - need to complete the integration of their information systems before they can develop a comprehensive data warehouse. Without mergers, some local service providers have active or residual differences in their data structures for the variety of jurisdictions they serve - the result of regulatory requirements.

Organizational boundaries also can complicate the original development of the warehouse design and present ongoing conflicts concerning ownership of the new transactions necessary to update the database. Valuable data can come from external sources, such as market research or demographic information. This can add substantially to the cost of maintaining the warehouse and the update mechanism.

Hardware/operating software. Selecting hardware and operating software for the warehouse involves both planning and speculation. There is almost no limit to how much and what kind of data can reside in the warehouse, and electronic storage is relatively cheap. It is tempting for management to imagine warehousing every line from every bill since Bell phoned Watson.

In execution, though, maintaining unneeded data is costly, and an overly large database can slow response times for users. One approach is to start on a slightly smaller scale, resolve the usability issues, demonstrate success (always useful for gaining new funding) and add to the database with a better knowledge of how the warehouse really will be used. For this reason, one important attribute of the hardware should be scalability.

Outside systems integrators often have specialized skills to establish a data warehouse. Dozens of reputable software companies have made significant commitments to developing and installing warehouse products, including applications tailored for telecommunications providers. Still, most service providers' warehouses are established in-house. Some of these decisions are based on long-standing corporate culture rather than painstaking analysis.

Analytical tools. The software developers that provide the tools to interrogate the database often assist companies with selecting the proper statistical tool for each business question. One significant difference between warehouses and relational databases is that most relational databases can summarize and report only from a known structure. Statistical tools can work at a higher level, generating the summary and the information request itself.

These tools can review a series of variables and identify unexpected relationships they find among the data. A relational database can answer the question, "Which customers had more than $300 in monthly usage last year?" A warehouse can ascertain, "What characteristics are common to customers who had $300 in usage?" It is then the marketer's turn to ask, "Who else has those characteristics?" (Table 1).

Success factors

Warehouses are expensive to build and expensive to maintain. Moreover, most current warehouse owners still consider it too soon to confirm the benefits of maintaining a warehouse. For a warehouse to justify its expense, it must meet the following criteria:

- Warehouses need to identify the non-intuitive relationship - this is called the "duh" hurdle. If the warehouse uncovers that "families with higher income are likely to own more computers than others," it has not met its burden. The benefit of statistics coupled with technology is to find the data relationships that are important, yet buried in the data.

- The answers must be timely. Warehouses are useful when they identify emerging trends, not when they document historical fact. In a market that changes at the speed of optical fiber, the earlier the trend reveals itself, the better. To benefit from this knowledge, the warehouse owner might need to develop a new service, upgrade its infrastructure or plan a marketing campaign.

- The results must be worth pursuing. A warehouse that identifies a statistically significant market of 10 prospects is not earning its keep. The warehouse analysis needs to rank the opportunities it identifies in a variety of ways to maximize their positive impact.

- System performance must excel. Ignoring an existing data warehouse is not as visible to management as, for example, ignoring the payroll system. For users to derive benefits, the warehouse must meet users' requirements for response time, usability and value (Table 2).

Pitfalls

Companies are racing to build warehouses with an admirable confidence, yet no telecommunications provider has offered proof that the warehouse has inflated its bottom line, and no astute provider will squander its strategic advantage when the benefits of its warehouse are quantifiable. Still, those contemplating making an investment in a data warehouse need to be aware of the risks and unresolved problems.

Warehouses combine the biggest new-systems development challenges with a company's largest repository of legacy data - all to be operated by non-technical users. Building a warehouse is, at its smallest, a large undertaking. It combines data from one end of the organization with data from other sources in and out of the organization. Most users will be unfamiliar with the IT and the statistical methodologies. The multitude of stakeholders who provide information or use the results could expand the scope of the original project and have a commensurate increase in cost, development time and complexity.

While an average-sized warehouse will cost millions of dollars, measuring its benefits will not be easy. Normal information systems efforts can predict and document reductions in labor, improvements in service quality or other tangible benefits. Warehouse efforts predict revenue increases or cost reductions that are impossible to specify prior to implementation and are not much easier to identify after achieving them. What status report could diligently itemize bills that would not have been paid, churn that did not occur or new markets that would not have been detected?

Active management commitment is essential. All interdepartmental efforts call upon management to resolve disputes and continue to motivate team members. Training, testing, documentation and other requirements that are orphans of so many information systems projects must be at the forefront of a warehouse implementation. If they are not employed, the warehouse is useless. Management must be prepared to make a financial commitment and an active role in the planning and the rollout.

In the end, the decision to build a warehouse might become a moot point for carriers. In a fiercely competitive telecommunications environment, the investment that one carrier makes to improve its competitive position will be required of all its competitors. The operation of data warehouses, as did automation in general, could become a necessity for a business to compete.

To achieve a competitive advantage, telecommunications providers undoubtedly will need to excel in the execution - not the simple existence or size - of customer profiling. The areas of opportunity include the applications of analytical tools, user commitment to its operations and, most important, the ability of the provider to make valid decisions and to act quickly on its discoveries. These are cultural requirements that demand more than simply a warehouse full of money. Monopolists are unable to accomplish these tasks. Fortunately, former monopolists turned into marketers should have no trouble.