By Philip Johnson
As a predictive analytics company serving health plans, Decision Point Healthcare Solutions has access to a wealth of data on health plan members. One of Decision Point’s biggest challenges is taking all that information and using it to actually change member behavior. I recently sat down with our linguistics team to discuss how Decision Point translates data into words that in turn can be used to motivate members to engage.
Tell us about the team’s linguistics background and why it is pertinent to Decision Point.
As experienced language professionals, we look for the balance between the art and the science in all that we do. At Decision Point our focus is revealing linguistic generalizations in daily life. As we work with different health plans and the segments they need to reach, our job is to come up with the content that will guide plan members through a process. We listen to member responses and develop a sense for how they will react. Just adding or removing a single word, like soon, now, or just, can drastically influence the response or action.
Is there a template for these communications?
At Decision Point, data-driven segmentation drives member engagement. With data at the heart of what we do, we can eliminate the guesswork and get to the goal. Decision Point’s models predict actions related to specific clinical measures or a member’s likely behavior. Then we use the drivers to group members into different segments, which are then treated as a community. It’s a basic three-step process where Decision Point:
- Takes all inputs and identifies members at risk of not taking actions or leaving the plan;
- Identifies drivers of activity or non-activity (desired outcome);
- Decides what content to use as triggers to engage the member and lead them to a path of action.
Composing the communications is an art, almost like writing a twitter message or a healthy haiku. We want to whisper a little suggestion in their ear. The communications follow a simple pattern. We greet the member warmly, ease into the conversation, engage with segment driven messaging, succinctly deliver the call to action, then close with encouragement and an offer of further support.
Can you give us some examples of member segments and the corresponding messages?
Some of the most obvious at-risk situations that we try to identify and influence revolve around members relationships with their primary care physician (PCP) and how that affects their health. We want to understand the drivers or lack of drivers for these situations, then address the situation with an approach that resonates with members. Here are a couple examples:
Poor relationship with PCP — “Everyone deserves to have a doctor they work well with.” The goal is to learn more about the member and to guide them in with follow-up messages, such as, “We’d like to help you find a doctor you can work well with.”
Multi conditions vs a chronic condition — “Managing your health can be a lot of work. We understand and are here to help.” We want to make them feel that they are not alone and that they are not the only person unsure of the next move. In this scenario, the segmentation is driven by complexity. These are people who are sick and overwhelmed by everything. They are wrestling with clinical complexity, comorbidity, and lots of events. They may be seeing specialists for several diseases. The goal could be to get them into care management and the call to action would be: “Many of our members have unique health challenges, and we want to help you with yours.”
Unengaged – These members are reluctant or just too busy to go see the doctor. So we send a message that says: “We know that getting the appointments and care you need can be tricky so we are calling to help.” Here the call to action can simply be help setting up an appointment.
The takeaways are similar with each segment. We’re moving members to their next step of engagement. It’s how we address the segments that draws the members in. We do this with just a single sentence. The sentences are quite short and flow naturally so that it is not obvious that there is specific targeting.
How does Decision Point measure the effectiveness of these campaigns?
We measure the effectiveness of how a message is working in conjunction with a control group for each segment. We look at which messages worked and which did not. Then we do some fine-tuning. Overall, the results depend on what we are trying to influence. For instance, retention is measured differently for each type of plan:
- Medicare Advantage Plans — members renew on an annual basis between October and December. In January, plans can look at their updated membership files and see whether or not they were able to influence member retention.
- Medicaid Plans — Annual renewal is based on the member’s original enrollment date. It’s easier to make calls 60 or 90 days out, then measure the impact three months later.
Does the message vary for different channels?
How language varies by channel is quite important, too. We tailor the message for text, email, live call or automated call (IVR) communications. The words matter and they also matter according to channel.
A text message might not work for an IVR outreach. Text must be super succinct. You can’t send five in a row. For recorded messages, you want to use more informal language that exudes natural warmth. We also want to address people at the appropriate reading level/understanding level, while not sounding condescending. It should be comfortable. This is especially true with Medicaid subscribers where the messages must be on a fourth or fifth-grade level.
Ultimately we try to compose non-threatening messages that reach members where they are. They leverage data driven segmentation and target specific risk tiers while being both warm and empathetic to the members’ challenges. When it’s done right, the communication feels natural and engaging.
Philip Johnson is Senior Business Analyst at Decision Point Healthcare Solutions