Analytics & Big Data
Dec 14, 2016

Improving physician engagement through multi-channel data: Physician segmentation

Among all the analytics buzzwords, "segmentation" is probably the most overused and under-defined. Segmentation means, essentially, the grouping of entities. In the simplest form, these groupings could be females vs. males, or kids vs. youngsters vs. adults. In more advanced form, segments are developed out of a population split via business rules or statistical derivation of virtually any quantitative and quantifiable metrics.

As the marketing world has progressed from one channel to multiple channels, the quantity of data available for analysis has increased immensely. A plethora of segmentation schemes can be easily developed, some from non-channel data and others from either raw multi-channel data or insights from multi-channel analysis. With a dozen of physician segmentation schemes in front of you, an immediate question is: “How useful are they? What actions can be taken?"

Segmentation output is expected to drive insights or actions or both. For example, physician-profile segmentation is typically for insights while promotion-response segmentation is for action. An ideal segmentation scheme should support both insights and actionability. When placed on the evolution path of marketing, physician segmentation is observed to have progressed through three stages mentioned in the below figure:

Figure: Physician Segmentation

Stage 1

Before the era of multi-channel marketing, physician segmentation leveraged physician profile data such as age, years of practice, and other demographic information. It answered the question, “Who are our physicians?" While profile segmentation neatly summarizes information about physicians, it has the following problems:

  • It does not align with a clear business question
  • It is ideal for the purpose of description and understanding, yet not for action
  • It puts excessive interest in physician identities instead of product features that matter most to physicians
  • It does not address physician behaviors that reveal attitudes and relate to business outcomes

Stage 2

While Stage 1 segmentation is disconnected from targeting, Stage 2 segmentation is more behavior-driven by incorporating product-related features or metrics. For example, including scripts written for drugs sheds light on a physician's preference. In contrast to Stage 1 segmentation which answers the “who" question, Stage 2 segmentation answers this question, “Which brand—or what brands—does each physician like to write?" Physicians holding a set of defining features for a segment have predictable business outcome, which makes this kind of segmentation more actionable. Stage 2 segmentation thrilled marketers when it first appeared as a behavior-oriented segmentation solution. Knowing which products your customers like is very conducive to targeting. Implementation was easy since there were just a few channels. With the surge of emerging digital channels, however, people realize the gap needs to be filled in order to reach the desired business outcome. This is the “how" question.

Stage 3

The question addressed by Stage 3 segmentation is, “How to drive physicians to write?" This is where multi-channel analysis becomes relevant. Multi-channel analysis gives the impact of each channel on sales for each physician, and the resulting segments each show a distinctive response pattern. Marketing tactics can be developed according to the channel preference pattern displayed.

Physician segmentation derived from multi-channel data is completely actionable and creates a bridge between customers and marketing strategies. Segmentation projects have long been known to fail at implementation for various reasons. Multi-channel-based segmentation, however, directly ties to marketing channel responsiveness and preference. It is straightforward to act upon and easy to measure.

Stage 3 segmentation can be further enriched by physician profile information as a view to the total marketable universe. It is descriptive, insightful, and actionable.

In conclusion, companies should conduct self-evaluations to decide where they are, and where they want to be, on this segmentation journey. Specific forms of segmentation may vary, but it is desirable to orchestrate segmentation schemes in such a way that they allow action plans to be developed and linked to business objectives.

About the author

Jingfen Zhu

Jingfen Zhu

Chief Scientist, Analytics, Chief Science Officer

Follow Jingfen Zhu on LinkedIn