Chatbots and health benefits


Chatbots could help humans take advantage of their health benefits

Conversations improve member understanding.

Anita Lee May 11, 2017

People acquire health insurance primarily to protect themselves from expense when they’re unwell. Viewed through this lens, standard coverages like doctor visits, emergency room care and prescription coverage are the most important parts of an insurance policy. But there is a completely different class of underutilized benefits that can have more impact in the long run. These benefits, typically known as “wellness programs” are tailored to help people avoid the health problems that can result from high-risk health behaviors like tobacco use and unhealthy diet. In addition to helping limit high-risk behaviors, these programs frequently incorporate preventative care like exercise and mental health programs too. Their goal is to change a consumer’s behavior to avoid becoming unwell in the first place.

These valuable benefits—from smoking cessation kits, to gym membership discounts, to voluntary vaccinations, and counseling services—are often underrepresented in promotional material or member management tools. Insurance companies are willing to pay for these cost-saving programs because they are valuable to the provider’s bottom line, but sometimes wellness benefits aren’t front and center because they can be challenging to administer across an entire population. They require the healthcare consumer to engage proactively. We see three user experience barriers to adoption that are burdensome for the provider:

1. The consumer need to learn what benefits are offered
2. Selecting a benefit that is personally relevant
3. Taking action to redeem a benefit (like going to the gym or getting a vaccine)

Opportunities to make big gains from small improvements

There are a number of ways that insurance companies can better communicate the value that these programs provide to their members to increase enrollment. The improvements don’t have to be expensive, complicated custom solutions. Proactively reaching out to members about relevant wellness programs and encouraging members to think about their daily health more often are great places to start. Later in this piece, we’ll describe an approach that uses conversational technology to address this low hanging fruit. But before we get there, let’s get some background.

The impact of preventable care

In 2006, American hospitals spent $30.8 billion treating preventable diseases. For context, that’s one of every ten dollars of hospital spending. While quality and access to treatment are major factors in keeping these diseases “preventable,” millions of potential hospitalizations could be avoided each year with increased adoption of healthy behavior at the individual level.

For insurance companies, the incentive for offering these benefits is clear: By improving the average health of their population, they pay less for coverage in aggregate while also improving the lives of their members. Insurance payers continuously develop benefits and wellness programs since participation in these programs can result in measurable quality of life improvements with lower care costs.

After looking at benefits offered by top health insurers like UnitedHealth, Humana, Aetna, Cigna, and Blue Cross Blue Shield, we noticed opportunity areas that could optimize the benefits and wellness programs for both the payers and members. On one extreme, benefits and wellness programs are displayed as “wellness marketplaces” where all of the searching and selection is left to the user—they have to determine which of the dozens of products and services are relevant, if any. On the other end, some payers ask users to fill out lengthy, time-consuming surveys to get a more tailored suggestion.

Neither approach makes much sense. Members who are so motivated that they log on to their insurance plan’s website, find the benefits marketplace, and search for the specific benefits that are most relevant to them may enjoy the additional wellness benefit, but they probably don’t need prompting. And people who are looking to improve their wellness but don’t know where to start are probably going to be discouraged when asked to fill out a 30-question survey that highlights how much work they have to do to get healthy.

The right channel

“We want to make the best for the most for the least.”
–Charles and Ray Eames

We thought about the advantages a conversational interface could contribute in fixing these issues (similar to how we thought about them in relation to retail banking products). This approach could offer wellness services two-way communication, asynchronous notifications, and would be clear and direct. A conversational interface that does not require heavily designed components would be a great fit, and since the service is very simple, it could easily be driven by a bot (SMS, Facebook Messenger, web-based chat agents) for the vast majority of communication. And while there are obviously data responsibility and legal issues that would need to be resolved, like communication opt-in and HIPAA compliance, keeping the service extremely simple will make those issues much easier to resolve.

General conversation model
A general conversation model

Below are two scenarios that demonstrate the type of conversations that payers can initiate with members. The first addresses the most high-risk factors and draws attention to a specific wellness program. The second scenario shows how a payer could actively help members optimize their health plan usage by reminding them of unused benefits and new programs.

There are major design challenges that would need to be addressed were this to be implemented as a product—users’ discomfort with this kind of conversation with their insurer; channel security; and onboarding. However, this approach offers the major benefit of being relatively inexpensive relative to the potential benefit: If even a tiny fraction of members stop smoking or get the vaccine that saves them a hospital visit, the cost to cover that group will drop precipitously, and the health of the whole population will be improved.

Scenario One: Greg wants to stop smoking


The first time the service reaches out to the member, we reference his sign up date to assure him that the message is coming from a legitimate source (the insurer). From there we remind him that this conversation is meant to benefit him, and then promptly move on to ask about our highest priority health issue: tobacco use.


One of the primary indicators of success in smoking cessation is the smoker’s willingness to quit. If the member isn’t interested in quitting, both parties will save time by switching gears to the next pressing health issue.


We’ve learned that the member wants to quit smoking, so we immediately tell him that his plan gives him access to smoking cessation programs. These programs are sometimes hard for users to discover, so it’s important to use this opportunity to make him aware of these services. We want to provide the easiest point of entry into the smoking cessation program, maximizing the likelihood that the member will participate.


At this stage, he’s ready to begin the smoking cessation program. One of the most important factors when it comes to quitting smoking is outside support. In acknowledgment of this, the system reaches out to the member in a week to check on his progress and continues to make relevant, mutually beneficial suggestions.

Scenario Two: Sarah should get a vaccine


We start our conversation with Sarah in the same way since we don’t know much about her yet. In these scenarios, the default highest-priority issue is to encourage smoking cessation, thus all members are probed quickly about this issue.


After learning she isn’t a smoker, we move on to the next high-priority health topic: preventative care. Yearly check-ups are the highest recommended benefit. If we can encourage the member to meet with her doctor in person, the doctor would likely be an advocate for these same types of preventative measures, too. A huge win.


In this case, the member has already had her yearly check-up, so we make sure she followed through on her doctor’s advice and got her flu shot. If she hasn’t, this low-effort goal could have a huge impact on her health. Dividing goals into quick, achievable actions can motivate members to finish one and move onto another.


We’ve learned that the member hasn’t had her flu shot this year. Since she’s met with her doctor for a check-up, there’s a chance that her doctor recommended a flu shot, but she chose not to get one. This is an opportunity to remind her of the personal inconveniences of getting the flu and a chance to re-iterate that the flu shot is covered by her insurance. We make sure that Sarah is comfortable with getting a flu shot while also providing a point of view.


The most immediate way to get a flu shot is to walk into a clinic like the Walgreens Advocate Clinic, so we show the member clinics in her area as a way to highlight how quick and easy it is. We also remind her that she can make an appointment with her primary doctor. A doctor’s visit is less immediate but ensures that she takes advantage of this benefit in an environment where she’s comfortable.



Just like with Greg, we follow up with Sarah after one week to see how she’s doing. In this case, she went out and got her flu shot, so she’s done everything she needs to do with respect to preventative care. After congratulating her, we start this conversation by telling her about gym membership discounts, moving her into the next health category: exercise and fitness.

Takeaway: Design principles

Wellness programs offer a unique opportunity for payers to incent positive behavior, improving members’ lives in the long-term and lower the cost of care for both parties. However, behavioral economics tells us that we are all really bad at using long-term outcomes to drive short-term decision-making. Here are a few design principles that can help in solving this or similar problems:

Match members with benefits that are effective for them rather than making them fish for their own.

Currently, insurers don’t offer a recommendation for which benefits would be most impactful at an individual level. Instead, benefits are presented in a menu, even though some products are more valuable for both the insurance company and the member. Insurance companies already collect data on program value and adoption rates, and using this data to make clear recommendations is actually a member benefit. For example, smoking cessation tools surely have a far greater return on investment than step counters, and a far greater potential to affect members’ lives positively.

Reach out to members directly, don’t wait for them to come to you.

A small percentage of members will proactively take advantage of available benefits, but relying on all members to take it on themselves to find wellness programs isn’t an effective strategy. Instead, reach out directly to remind members that their insurance plans have value in their everyday health, not just at high-impact moments.

Ask simple, direct questions to learn the most immediate health opportunities and interests.

We can’t make a good recommendation if we don’t know anything about our user. On the flip side, we also can’t let our need to collect information become a burden on the user. By focusing only on the most important topics, we can reduce long questionnaires into a simple and short Q&A.

This post was a collaborative effort between between Anita Lee, Steven Bennett, and Josh Lucas-Falk.

Anita lee

Anita Lee