Bayesian Analysis

UnitedHealth Group is implementing the Bayesian paradigm in marketing to enable data-driven decision-making that continuously improves based on customer feedback (Spiegelhalter et al., 2004). During enrollment for the Healthy Living Rewards pilot program, 100,000 members completed preference surveys identifying what they want from wellness programs and what improvements they wish their current providers would offer.

Members indicated current wellness providers lack:

1. Flexible Participation Options: Current programs require fixed schedules and locations

2. Digital Health Tools: Existing programs rely on paper-based tracking and in-person meetings

Analysis 1: Flexible Participation Preference

Prior Probability (Before Survey):

Industry research suggests 55% of wellness program members request flexible participation options (Robroek et al., 2009). This represents initial belief before collecting company-specific customer data.

P(Prefers Flexibility) = 0.55 or 55%

Survey Evidence:

Pilot member survey results:

- Members wanting flexibility improvements: 68,000 out of 100,000

- Members not prioritizing flexibility: 32,000 out of 100,000

P(Prefers Flexibility | Survey Data) = 68,000/100,000 = 0.68 or 68%

Bayesian Update:

Customer evidence reveals 68% prioritize flexibility, higher than 55% industry benchmark. The posterior probability of 68% replaces the 55% prior, representing updated belief based on actual customer feedback.

Identifying Similar Customers:

Parents with children show even higher flexibility preference, representing a key target segment:

- Total parents surveyed: 34,000

- Parents prioritizing flexibility: 29,580

P(Flexibility | Parent) = 29,580/34,000 = 0.87 or 87%

Interpretation: Members similar to this profile (working parents with childcare responsibilities) demonstrate 87% probability of wanting flexible wellness options, substantially higher than 68% overall average.

Analysis 2: Digital Health Tools Preference

Prior Probability (Before Survey):

National surveys show 35% of commercial insurance members use digital health platforms (Krebs & Duncan, 2015).

P(Prefers Digital Tools) = 0.35 or 35%

Survey Evidence:

Pilot member survey results:

- Members wanting digital tool improvements: 59,000 out of 100,000

- Members not prioritizing digital tools: 41,000 out of 100,000

P(Prefers Digital Tools | Survey Data) = 59,000/100,000 = 0.59 or 59%

Bayesian Update:

Customer evidence shows 59% prioritize digital tools, significantly exceeding 35% national baseline. This 24 percentage point increase indicates UnitedHealth's target customers want substantially more digital engagement than current market offerings provide.

Identifying Similar Customers:

Younger members ages 25-44 show highest digital tool preference:

- Total ages 25-44: 62,000

- Prioritizing digital improvements: 47,120

P(Digital Tools | Ages 25-44) = 47,120/62,000 = 0.76 or 76%

Interpretation: Members similar to ages 25-44 demographic demonstrate 76% probability of prioritizing digital tools, indicating this segment strongly desires mobile apps and online engagement currently missing from traditional wellness programs.

Combined Analysis:

Some members want both improvements simultaneously:

P(Both Flexibility AND Digital) = 41,000/100,000 = 0.41 or 41%

Working parents ages 30-45 represent the highest-value segment wanting both features, creating premium program design opportunities. These demographic segments overlap, most parents fall within the 25-44 age range thereby enabling precise targeting based on combined characteristics

rather than isolated attributes.

Visualization

Figure 1: Bayesian Probability Update: Prior Vs. Posterior

Figure 1 illustrates how customer evidence updated initial beliefs. Flexibility preference increased from 55% prior to 68% posterior (13 percentage point increase). Digital tool preference increased from 35% prior to 59% posterior (24 percentage point increase), demonstrating customers want digital improvements even more strongly than flexibility enhancements.

Recommendation: Implementing Bayesian Marketing Paradigm

Based on this analysis, I recommend UnitedHealth Group fully implement Bayesian marketing methods across all customer engagement initiatives for three strategic reasons:

1. Continuous Learning from Customer Feedback

Traditional marketing relies on static annual surveys producing outdated insights. Bayesian methods continuously update beliefs as new customer data emerges. As additional members provide feedback, probabilities refine automatically. After 25,000 additional enrollees, updated posterior probabilities become the new priors for subsequent analysis, creating self-improving customer understanding.

2. Personalized Segmentation Using Similarity

Rather than generic communications, Bayesian analysis identifies customer segments with similar preferences. Parents receive flexibility-focused messaging (87% preference probability), while ages 25-44 receive digital-tool-focused messaging (76% preference probability). This precision targeting increases engagement by delivering relevant features to each segment.

3. Data-Driven Resource Allocation

Bayesian probabilities quantify customer priorities, enabling evidence-based investment decisions. The 59% digital tool preference (up from 35% prior) justifies allocating 60-70% of technology budget to mobile app development rather than traditional materials. The 68% flexibility preference supports implementing on-demand video classes and virtual coaching over fixed-schedule group sessions.

Note: Key Segments Identified:

- Working parents ages 30-45: Highest engagement probability (want both features)

- Ages 25-44 non-parents: Strong digital preference (76%)

- Ages 45-64 parents: Strong flexibility preference but lower digital (62%)