Using a variety of statistical methods, including decision tree algorithms, self-organizing maps, and latent class analysis, marketers can marry the responses with the physicians' demographic data and prescribing behaviors and identify the variables that cause greatest differentiation. Out of dozens of variables, typically the top four or five variables as identified by the segmentation technique determine the segment physicians will fall into. As one might expect, that segmentation differs from product to product, depending on the characteristics of the therapeutic category and the drug in question.

During the analysis phase, executives profile and name the segments to begin building a predictive model. Some physicians might fall into a segment described as "product innovator" and "rep friendly," others into "product conservative" and "rep unfriendly." (See "A Closer Look," page 70.) Once the segments are defined, secondary information, such as prescribing behavior data, managed care influences, and group practice membership help shape the predictive model that projects the segments all doctors within the target market will fall into. And once companies assign physicians to segments, they can automatically deduce their psychographics and attitudes because the segments were initially developed with primary, representative research on those variables.

Analysis is Not Enough Definition and analsis of segments, and the ability to project psychographics and attitudes, are meaningless unless they are used to transform a pharma company's marketing and sales processes.

Factoring in all such data, pharma executives can develop marketing and sales approaches for each segment. During that process, marketers need to take a hard look at all current promotional materials. Some may still be usable, while new collateral material may need to be created. Marketers may even determine that sales reps shouldn't visit some segments, even those that contain high prescribers, if they are unreceptive to detailing or if the possibility of those physicians prescribing the company's drug is very low. As an alternative to an in-person detail, companies may find that using direct mail or e-detailing is more productive and cost-efficient for those groups.

For segments that sales reps should concentrate on, the promotional plan will indicate

• detail frequency
• messages to support the product and messages to counter physicians' objections
• what materials to exclude
• what other promotional tools-such as e-letters, thought leadership and peer influence programs, and programs for fellows/clinical liaison and medical information scientists, should supplement the detail.

"The result of the segmentation process can't be academic," says UCB Pharma's Banet. "If you use the word 'psychographic' with a sales rep, he or she will immediately tune you out. They want to know how to be more effective. Is this physician interested in clinical data? Is he or she influenced by peers?

"The final product must have real-world implications for interacting with doctors. When a rep identifies a doctor as a 'segment number six,' that rep should immediately be able to paint a picture of what that doctor's characteristics and needs are and be able to customize their message accordingly. At the end of the day, we need something that's not only strategic, but tactical and implementable."

Piloting Pays Off Before completely rolling out such a program, companies should implement a pilot involving a small percentage of the sales staff. It's important to ensure that participating reps are appropriately compensated, no matter how the effort turns out.

To make the assessment valid, executives must carefully choose test and control territories. Companies that quantify the effectiveness of the integrated segmentation approach by tracking and measuring results help gain assurance that it works before the full roll-out. In addition, assuming the results are positive, the rest of the sales force subsequently will be more willing to adopt the concept when companies implement it companywide.

To further strengthen sales rep buy-in, marketers should share how they developed the segments and administer exercises that engage reps in the process. Role playing, in which reps interact with mock physicians, can also teach reps how to best employ the recommended strategies and counter physician arguments against product positioning. No predictive model is 100 percent accurate. However, integrated segmentation can predict the correct segment for any given doctor with 75 80 percent accuracy.

For the rest, sales reps are armed with one or two questions to ask physicians or their staff that will tease out the appropriate segment membership. The response to "How concerned are you about managing your patients' pain?," for example, can confirm whether a doctor has been accurately placed in a segment or whether he or she should be moved into an adjacent one.

Finally, no pharma company using an integrated segmentation approach can afford to be satisfied with a single study at one point in time. Doctors and markets are constantly evolving. Marketers should review segments and the physicians within them annually or after every major new development.