 |
Most marketers are familiar with geodemographic segmentation, where
small geographies such as census tracts, zip codes or neighborhoods
are clustered into relatively homogeneous groups with similar demographic
profiles. Commercial geodemographic segmentation systems such as PRIZM,
MOSAIC and ACORN, have been around for 30 years or so, and typically
contain thousands of neighborhoods across the United States grouped
in 50 to 60 clusters with suggestive names like “Enterprising
Couples,” “Dream Weavers” and “Upscale Suburbanites.”
The basic idea behind these systems is that “birds of a feather
flock together” – so that marketing efforts tailored to
the dominant demographic characteristics in an area can be effective.
While popular for direct marketing to consumers, these systems are often
criticized because their exclusive focus on demographics and lifestyle
factors ignores many other important attitudinal and behavioral factors
relevant to consumer purchase decisions.
Since most geodemographic segmentation systems focus on overall general
population characteristics rather than populations at risk for or diagnosed
with a specific disease, they are also not typically used for patient
segmentation. Nevertheless, both geography and demographics are important
factors in pharmaceutical marketing. The prevalence of many diseases
correlates strongly with demographics such as race/ethnicity and age,
e.g., HIV, diabetes and Alzheimer’s disease. Physician prescribing
patterns and patient brand preferences differ across geographic regions.
Pharmaceutical sales forces are deployed and managed on a geographical
basis. Pricing and reimbursement practices vary significantly across
states.
This article, authored by Doug Willson, Ph.D., Senior Vice President,
Marketing Science at GfK Strategic Marketing, describes how to combine
primary survey research, secondary data and GIS (Geographic Information
Systems) to develop and deploy actionable geodemographic segmentation
solutions in pharmaceutical markets. Fully featured GIS are now available
at very low cost as desktop or web-server applications. Applications
that integrate patient-level segmentation and GIS can now be developed
as stand-alone tools or tightly integrated with existing customer relationship
management (CRM) systems. When combined with primary research and secondary
data sources, GIS can support many different strategic and tactical
marketing applications.
Geodemographic segmentation research typically begins with a preliminary
analysis of secondary data, including census information, epidemiological
data from federal sources, anonymous patient data, physician prescribing
data, and perhaps other secondary sources. Ideally, the preliminary
analysis helps develop an understanding of the geodemographic correlates
of the specific disease and allows us to size patient populations
in small geographic areas based on the demographics of the region. At
this stage, it is tempting to mirror the traditional geodemographic
segmentation approach and develop neighborhood segments directly from
the secondary data. However, just as in the traditional case, there
is no guarantee that pure geodemographic patient segments will be useful
for marketing purposes; it is better to proceed to primary survey research
with patients, using the information from the preliminary secondary
data analysis to design a sample that provides both geographic and demographic
representation of the relevant patient population.
The patient survey typically investigates the patient’s disease
state, current treatment and treatment history, attitudes toward the
patient’s illness, treatment preferences, unmet needs, and the
relationship with the patient’s physician. The survey information,
in combination with the geodemographic information from the first stage,
can then be used to create segments using multivariate clustering methods
(like K-means or latent class). As in any segmentation, the selected
segmentation solution will ultimately depend on the goals of the segmentation
and the specific strategic and tactical applications that it will support.
For example, in HIV, some potential segments might focus on diagnosed
but untreated patients; in this instance, segment-specific communication
strategies might be designed to educate patients about treatment options
and the benefits of earlier treatment, and to encourage them to seek
treatment.
Once the patient segments have been developed, statistical modeling
is used to “score” the geographic database underlying the
GIS. This is a very different activity than scoring a database of individual
physicians or patients. In the GIS, the lowest geographic layer (i.e.,
census tracts or zip codes) is scored with an estimate of the number
of patients that fall into each segment. After the segment scores have
been included in the GIS, customized maps and tables can be developed
for any geography of interest – states, sales districts and territories,
market areas, or physician-specific trade areas.
Perhaps the simplest example of a tactical segmentation application
supported by GIS would be a classical underpenetration analysis
of geographic market areas such as sales territories, cities or MSAs,
or states. In this analysis, the number of patients within each segment
is estimated for each market area, and each brand’s share of prescriptions
is calculated and compared with an average or desired level given other
characteristics of the area and the patient population. Areas that are
underpenetrated (and perhaps should be targeted for marketing interventions)
are those where a brand’s share of prescriptions is less than
the desired or average level. In the example below, market areas in
Pennsylvania that are underpenetrated for a particular brand are emphasized
in dark red.
Analysis of physician-specific trade areas provides another application
of GIS and patient segmentation. Physicians and hospitals currently
perform trade area analyses when choosing office and hospital locations,
and when examining their own competitive situations vis-à-vis
competitors after their site has been chosen. Trade areas can be defined
as a circle of specified radius, or in drive-time minutes, around the
office location. The appropriate definition of a physician trade area
will vary by disease category and specialty; we typically investigate
travel distance and drive time in the patient survey research. The example
presented below displays the drive-time boundary, the distribution of
patient segments within the market area, and a performance index showing
whether each brand is under or overperforming in the market area, relative
to potential. This analysis can be modified to account for adjacent
physician offices and overlapping market areas and used to support marketing
interventions at the physician level. It is important to note that the
drive-time analysis requires the “active” GIS – with
the detailed streets file and calculations included in the system.
Overall, there are many strategic and tactical marketing activities
that
can be supported through GIS and patient-level segmentation. Examples
include:
- Identifying and prioritizing geographic market areas by overall
opportunity
- Tailoring media plans to reflect the demographic makeup of patients
in
specific market areas
- More closely aligning sales resources with opportunities
- Developing and delivering segment-specific messages to patients
- Creating targeted patient education materials for local distribution
- Segmenting physicians using the characteristics of the patients
they
treat
- Developing messages for physicians that highlight the unmet needs
of
their patients, identifying those patients who will benefit most
from
specific therapies
|
|
