How behavioral science, segmentation, and hyper-personalization helps health plans activate their members.

For the past few years, the health care industry has responded to the rising demands of consumerism, resulting in a wealth of patient-centered strategies and applications, all designed to combat the historic lack of billing transparency and patient agency in health care. Many providers have embraced the shift by implementing a variety of digital self-service tools and patient-engagement initiatives—but payers have been slower to climb on board.

While the pandemic sparked rapid changes in care delivery nationwide, health plans have continued a trend of low consumer-satisfaction rates. In its most recent U.S. Commercial Member Health Plan Study, J.D. Power found that more than one-third of surveyed members (37 percent) and nearly half of surveyed members age 57 and older (44 percent) reported no engagement with their health plan.

This lack of engagement has a particular impact on Medicare and Medicaid markets.  Payers certainly understand that engaging these members is critical to optimizing risk adjustment, improving quality ratings, and driving better access and affordability. Yet precisely identifying and activating members—and driving the kind of behavioral change that improves health outcomes—remains an insurmountable struggle for many plans, especially when members vary significantly in age, ability, and need.

Harnessing AI and behavioral science to drive engagement

Rapid advances in behavioral science and artificial intelligence (AI) offer payers new opportunities to drive member engagement. By leveraging AI and data analytics to develop detailed knowledge about members’ needs, motivations, attitudes, barriers, and enablers, payers can break down traditional identification and activation barriers to achieve better outcomes and lower costs.

When payers take the time to define the disparate member segments that exist within their overall population, they can effectively tailor communication type, sequencing, and content to accomplish their outreach goals. In today’s environment, people are continuously besieged with messaging, even in health care; this is the era of information overload. Personalized communications cut through the noise by helping the recipient feel known and understood—and therefore much more likely to respond or participate.

But how does the behavioral science behind personalization work? Let’s use a health plan attempting to identify and enroll its dual-eligible population—those members who qualify for full or partial coverage with both Medicare Advantage and Medicaid—as an example.

Leveraging AI to identify target populations

Rather than simply cold-calling all its members over a certain age, the health plan relies on an AI predictive model to identify members who are most likely to be dual-eligible. The AI algorithm supplements the plan’s own demographic data with third-party demographic, lifestyle, and social determinants of health (SDoH) data to identify prime targets for dual eligibility. Some AI models can predict with up to 93 percent accuracy those individuals with the highest likelihood to qualify for full or partial dual coverage.1

After identifying its target population, the health plan uses empathy interviewing and the principles of behavioral science to create a hyper-personalization strategy. By surveying more than 1,000 members over the age of 55, the health plan now has a statistically representative sample size from which to draw conclusions. The survey asks more than 50 questions on a wide range of topics, including marital status, number of children, access to various technologies, military service, health status, socialization and support network, and contact preference.

Determining member segments and personas

The resulting data points are run through a cluster analysis to find naturally occurring groupings of members who share common characteristics. These groupings extend beyond the typical demographic delineations to include psychosocial factors, personal motivations, and attitudes toward technology. The health plan validates and compresses these groupings to determine seven key segments that define its 55+ populations.

For example, one segment is personified as the “Grandchild-Focused Matriarch,” a widowed or divorced renter with higher health care utilization and costs who is independent, religious, and motivated primarily by staying connected to grandchildren. Understanding member segmentation is key to determining which communication channel is most likely to be effective with which segment, as well as the kind of content—including language, imagery, and level of detail—that will be most persuasive.

Effective member segmentation depends upon extensive data analysis and applying a behavioral science lens to the results. For example, while the Grandchild-Focused Matriarch is likely to self-report a preference for phone calls rather than emails, the survey data revealed complicating factors: These members typically do not answer calls when they don’t recognize the number, and they use their laptops every day. As a result, the recommended order of contact is first a primer email, then a follow-up call, and then a text reminder.

Sometimes, health plans discover that their assumptions about a particular population group are misguided or too simplistic. The plan’s research revealed that its members on disability diverged into two distinct categories: one group of socially active and well-supported members who were highly tech-savvy, and one group of socially isolated members who struggled with lifestyle factors like transportation and had poor technological fluency. While the health plan could effectively engage the first group by text and email, the second group would respond much better to outreach from a live advocate.

Discovering trends and opportunities

This type of behavioral science research also reveals large-scale trends among a health plan’s targeted populations, which can result in significant strategy shifts. By understanding that only two of its seven member segments are likely to be motivated by mail-based messaging, for example, the health plan can eliminate the resource and administrative costs of mailing postcards to its entire over-55 population.

Behavioral science research can document and anticipate how the content and ordering of messaging impacts the outcomes a payer receives. For example, one health plan struggled to reduce the number of initial hang-ups it received from cold-calling its dual-eligible population. After many rounds of empathy interviewing, the plan realized that its members needed reassurance that talking about their health care status with a stranger was a good choice.

As a result, before initiating phone contact, the plan began sending members an email that explained that a live advocate would be calling them from a particular number. This ‘primer email’ detailed why the phone call was important for the member and described in advance the kinds of questions that would be asked. After implementing its primer-email strategy, the plan realized a 400 percent increase in members answering the phone and engaging with the advocate.2 Most surprisingly, this increased engagement persisted throughout the entire dual-eligible enrollment process, resulting in a higher number of enrollees.3 By anticipating and providing the transparency members needed, the health plan earned its members’ trust—and its efforts paid off.

Pursuing a sustainable engagement strategy

Pairing AI with behavioral science research amplifies the precision with which health plans can target members for a particular purpose, whether that’s enrollment in Medicare and Medicaid or encouraging those with chronic conditions to improve their medication adherence. By taking the time to understand the specific factors that motivate members to act, health plans can apply small but meaningful changes to their content and delivery that can significantly improve long-term engagement.

In every member category—even a seemingly uniform one, such as members on disability—there are always differentiators, and those differentiators are important. Channel and outreach hyper-personalization strategies are effective in helping members feel heard, seen, and cared for—and their resulting behavioral changes will help health plans reap the benefits of this new kind of member engagement.

1Change Healthcare internal statistics reflecting Dual Enrollment Advocate™ customer use of its proprietary, predictive AI model. Results may vary by payer plan and member demographics.

2Change Healthcare internal statistics reflecting more than 20,000 emails sent to more than 445,000 members in three beta health plan clients. Results may vary by payer plan and member demographics.


About the Author

Brock Vestrum, CSPO, product manager for member engagement, Change Healthcare, manages the member engagement solution suite at Change Healthcare.

Brock oversees the design and delivery of health care-journey products, including enrollment, program assistance, and complete-care engagment. With more than 15 years in the industry, he has developed an approach that delivers front-end consumer engagement applications and back-end efficient architecture to deliver a targeted user experience to drive behavior change. Learn more about Brock on LinkedIn.