I tried asking Alexa this question at home and got no response. Perhaps, Alexa was giving me the silent treatment! I tried variations of the question and Alexa eventually responded to a more generic version of my original question by reading off the names of about twenty health plans from a website.
I was hoping for more!
As the Medicare Advantage (MA) Annual Enrollment Period (AEP) is fast approaching, let us imagine a future where Artificial Intelligence (AI) plays a much bigger role in the health plan sales process. For the purposes of this blog, I am defining AI broadly to encompass a wide range of techniques including machine learning, natural language processing, clustering and pattern matching, and generative AI. Here are three “modes” in which health plans could ramp up their use of AI in sales.
Voice Assistant Mode
This is the futuristic mode where seniors directly ask Alexa (or their voice assistant of choice) for a recommendation on the MA plan they should enroll in. If the senior has developed a high degree of trust in Alexa, she might already be aware of the senior’s healthcare diagnoses, needs and preferences or could be provided those through a conversation with the senior, or through a permissioned application programming interface (API) with an electronic health record (EHR). This would of course need to be done carefully with strong data privacy agreements in place.
Alexa could then match the senior’s diagnoses, needs and preferences with the products available on an MA marketplace, augmented by additional information not directly on the marketplace like customer feedback on social media platforms and plan quality and compliance data from the Centers for Medicare and Medicaid Services (CMS) website. For example, Alexa may go beyond the plan’s aggregate Star rating to investigate the plan’s performance on the subset of quality metrics most relevant for the senior’s diagnoses (e.g., quality metrics for diabetes). Where available, Alexa could also encourage the senior to consider a C-SNP (Chronic Conditions Special Needs Plan), which may be a better fit for the senior in some cases.
Further insights could be gleaned by interfacing with an out-of-pocket cost estimator for the senior that considers a range of healthcare needs scenarios for the senior over the next year and how the plan options would fare under those scenarios, giving the senior an analytically sound estimate of the likely total range of out-of-pocket costs associated with the top plan choices. For example, Alexa could interface (in an appropriately permissioned manner) with historical claims databases to run customized scenarios of costs given the senior’s age, conditions, and other parameters. This would collectively enable Alexa to come back to the senior with her top choices, with an explanation why those would be the right choices for the senior. For the senior, this has the potential to be a more scientific discovery process than traditional sales methods (cue prospect sitting across the kitchen table with a broker).
But could this futuristic scenario ever happen? For one, regulation will be key to determining how far the voice assistant could go in the kinds of sales processes described above. Would Alexa need a broker’s license? In all fifty states? What happens to brokers’ traditional roles if this mode takes over for a sizeable segment of seniors? How would MA plans need to reorient their channel strategies and distribution value propositions in response to innovations like these? How ready are we, as a society, to go in this direction? All questions that need to be addressed by the healthcare industry, governments, and society.
Plan Analytics Mode
A nearer-to-home set of applications could leverage AI to power health plan internal sales processes. Plans currently buy leads from third parties, which are used for marketing and outbound prospecting outreach. With AI, plans could “enrich” these leads, with more meaningful information gleaned from sources like proprietary research, social media platforms, and past win/loss and seller/broker data from the health plan’s prior years’ sales operations databases. As my colleague Erik Swanson, who leads our Data and Analytics group, mentions, there is already a robust market around AI-enabled personalized marketing and sales, including increasingly ubiquitous AI assistants that can help improve the sales cycle.
Collectively, these analyses can both enrich the leads by adding information about the prospects as well as help plans prioritize and direct these leads to the internal sales team members or external brokers most likely to close the sale. One could imagine a variation of these analyses and AI-powered business processes deployed for inbound sales opportunities as well. For seniors, this has the potential to be a more tailored approach that reduces “spam” outreach and eventually enables better choices, when carefully executed.
Broker Assist Mode
In yet another set of applications, health plans could bring the power of AI in service of their broker partners. Building on the plan assist mode ideas mentioned above, plans could “feed” brokers the leads that are best for them to work on, by cleaning, enriching and prioritizing leads. Furthermore, plans could use AI to suggest the right products, messaging and next best actions for the broker that are most likely to advance the prospect through the sales process. All of this could be highly tailored and personalized – for example, plans could channel targeted leads depending on the broker’s history of success with certain types of prospects, messaging could be different based on whether the prospect is just aging into Medicare, considering switching from traditional Medicare or considering switching from another MA plan.
Moreover, plans could provide custom-built AI assistants to brokers to make them more effective and their business processes easier to execute. For example, brokers could use an ambient AI platform during a prospect conversation which prompts ideas/suggestions real-time or post-meeting; if that feels uncomfortable to the prospect, broker could call in their report of a prospect meeting from their car and AI can suggest the next best action. My colleague Erik Swanson shares that ambient AI is already being used at the “bedside” in clinical settings and demonstrates improvement in provider empathy. In our context, ambient AI could similarly help brokers adjust their tone to match the nature of their conversation with the prospect.
In Closing
The three modes described above are not intended to be mutually exclusive or collectively exhaustive. While MA has been featured prominently, these ideas are applicable in other lines of business as well like the individual market. Further, in the employer group market, AI assistants could help internal sales professionals and external brokers draft responses to employer RFPs.
While it is easy to get excited by the sheer potential of the technologies, plans need to leverage AI in the context of a thoughtful distribution and enterprise strategy, so that their efforts are tailored to their business and stakeholder context and their initiatives are addressing real opportunities, rather than getting caught up in the hype cycle. Plans that push hard on this front also need to be cognizant of broader regulatory, business partner and societal acceptance considerations. For example, while the broker assist mode should be welcome, an autonomous voice assistant mode has the potential to disintermediate the broker and invite backlash from these valued business partners. How societal trust evolves with AI will also influence when the more futuristic ideas in this blog come to fruition. Exciting opportunities lie ahead for plans that can navigate these considerations and out-innovate competitors.
The author would like to thank Erik Swanson, senior vice president and leader of Kaufman Hall’s Data and Analytics group, for his valuable insights and comments for this blog.
