Customer Segmentation

The best performing businesses know their customers better than do their competitors. This is true in healthcare too. Whether you run a hospital or sell pharmaceutical products, the better you know your customers, the better you can bring value to them (and make money from them to be frank…).

The more nuanced your understanding of who your customers are, what pain points they have, how they prefer to have the pain point solved, the better you can create and deliver products and services that address their needs, meeting them where they are.

What is it & How it’s used

Customer segmentation groups customers with similar characteristics. Each segment tend to exhibit the similar needs, motivations, education, ways of consuming services, willingness to pay among other characteristics.

Based on these characteristics, business would develop different marketing plans, sales channels and service delivery modalities. E.g. young mobile native customers and seniors with visual impairment would have vastly different needs and preferences.

Within the healthcare space, in addition to the above generic business segments, there are other salient segments, including medical condition, payer class. E.g, diabetes patients have very different needs than do mental health patients; further diabetic patient with >11 A1C would have very different risk and service utilization profiles than would patients with a <7 A1C; in the US, employer health plans would pay more per unit of service than would Medicaid plans (low income).

This is why you will find hospital teams formed around particular specialties, surgery, cardiology, internal medicine; pharmaceutical companies run teams around particular diseases, which aligns R&D, go to market strategies. It is also very common to see teams created by payer class, i.e. commercial, Medicare (>65 year olds) and Medicaid.

How to create segments

Well-functioning businesses have know well how they create these segments, i.e. they know what the key characteristics are about their customers. While the construct of these segments are derived from invaluable experience, traditionally these tend to be qualitative and if quantitatively constructed, often at a very high level. As data volume and sophistication improve over time, businesses should devote more resources to further segment their customers to develop more nuanced understanding of their customers.

E.g. let’s traverse one of these segmentation approaches

  1. start with a typical segmentation along age
    1. 0-18 youth, growing
    1. 19-25 young adult, invincible, accident prone
    1. 26-40 adult, child birth, healthy mostly
    1. 41-64 mature, rising incidence of medical conditions
    1. 65+ retired, frail, cancer, assisted living
  2. For each of these age groups, gender is the next obvious split
  3. Presence of chronic illnesses may be the next, e.g. diabetes, asthma, heart failure.
  4. Over laying all that, you can break out payer class, i.e. commercial, Medicaid or Medicare.

As you can see, the initial five age groups rapidly expanded into hundreds of much more granular segments. How you service a 20 year old healthy male on a commercial plans would be very different from a 75 year old Medicare female patient with late stage cancer living in palliative care facility.

BUT it’s not practical to create a customer service or sales strategy for each and every one of these more granular segments. In practice, you would prune, focusing on the few cohorts with the highest volume of customers/patients that present the greatest variations in characteristics.

The pruning process can be done very simplistically using bivariate analyses, but sometimes it’s best to use more sophisticated techniques like random forest (RF) and/or cluster analyses to understand the most salient factors differentiating your customers. Especially the multivariate nature of these segments quickly overwhelm the human brain. Using RF approach would require a single target to find similar cohorts e.g. total health care cost, whereas cluster approaches incorporate more characteristics into the cohort creation.

I’m creating a new course on Udemy on how segmentation can be done in healthcare soon. Pls subscribe to stay tuned.

Applying the segments

Once these segments are pruned, and defined in ways that the business and operations teams can utilize, the analytics team would codify the segmentation logic so that the segments can be created automatically. These segments would permeate internal platforms such as SalesForce and call center workbenches and trigger differences in how sales and services provided.

Doing this in an automated fashion requires input data to be consistent, thus drifts in data format/quality should be monitored and factor into segmentation logic maintenance.

Particular to healthcare, segmentation done using measures that change often require more care. E.g. a patient’s A1C can change over time, thus changes to how a patient’s care is delivered should ensure changes happen smoothly, to maximize patient adoption and satisfaction.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s