Blog Post

What HealthJoy’s Data Actually Tells Us About How Members Make Care Decisions

The biggest misconception about "predictive" in benefits is that it means certainty, like a crystal ball that says this person is having surgery on the fourteenth. That’s not what it is and it shouldn’t be. What we’re doing is identifying members who are more likely to be heading toward a certain type of care, early enough to be useful, so we can offer guidance. 

·

·

8 min read

Care navigation tends to start when the member does, which means it’s reactive by nature. David Lawrence, Vice President, Clinical Product at HealthJoy, leads the company’s work on how we think about member signals, and he works closely on where cost and quality meet in care decisions. That puts him near a question reactive navigation can’t answer: what happens before a member ever searches? In this Q&A, David explains what our data tells us about how members make care decisions, what a signal is and isn’t, and what HR leaders should ask of any navigation vendor. 

How members decide

What problem were you trying to understand when you started looking at how members choose care?

I wanted to understand the signals that let us get ahead of these decisions and how we can most effectively guide members to the best care as early in the process as possible. The reason is simple: by the time a claim arrives, the care has already happened.

It matters for two reasons: 1. For employers, a small number of decisions can have an outsized impact on total spend and 2. For members, an intuitive, personalized experience makes the search and the decision process much easier.

What have you learned about how members make care decisions?

First, decisions are made faster than most people assume. The moment someone has an acute need and the moment they start deciding how to address it are effectively the same moment. And once they’ve picked a provider, most of what that episode will cost has already been decided because the facility and the physician drive the price.

The second thing is that lead time depends on the type of care. A scan comes together quickly. A planned surgery or an infusion has a longer runway. So there isn’t one window. There are several, and each one needs a different kind of help.

What surprised you the most when you first looked at how members make care decisions?

It was part structural and part behavioral.

On the structural side, the signal everyone assumes is useful, “this member saw a doctor,” tells you almost nothing. Nearly everyone sees a doctor. And by the time a claims-based trigger comes through, the member has already chosen the surgeon, the facility, and potentially scheduled the appointment.

On the behavioral side, a lot comes down to where a member stands in their plan year, whether they’ve met their deductible, and how long they’ve been enrolled. Those shape what a member is likely to do next as much as anything in their claims history. It isn’t surprising that cost is a driving factor for many people. What surprised me was how clearly and consistently it plays out in the data.

There’s a window between when a need becomes visible and when a member picks a provider. Why does it matter?

It’s the window where the cost and quality of care get decided, along with many downstream aspects of care. Once that decision is made, many aspects of the future care are locked in. If you want to change the outcome, you have to be present inside that window, with guidance, while the choice is still open.

Why is that window so hard for most navigation tools to see?

Because most navigation tools are built to react. They depend on a very specific signal from someone: a member who raises their hand, searches for care, or asks for help.

The other late signal is the claim. Claims are the most complete record of care we have, but they describe what’s already happened and they arrive on their own schedule, often weeks after the visit. A tool that relies on claims learns about the scan after the scan. To see the window, you have to notice the earlier, more nuanced signals, and you have to refresh that picture every day - not just once per quarter.

The real challenge is to be proactive and trusted, and to have high engagement even outside those specific care-seeking moments.

Signals

In your own words, how does HealthJoy define a “signal” and what kinds exist?

We define a signal as any piece of evidence that a member may be heading toward a care decision. Broadly, there are three kinds:

  1. What members tell us directly: a search in the app for a specialist or an imaging center, a question they ask us, or a health goal they set. 

  2. What the plan tells us through claims: a new diagnosis, a visit pattern, or a visit that was completed. 

  3. Context: where someone is in their plan year, what their plan looks like, how long they’ve been with us.

When taken together, we learn much more about which kind of care might be on the horizon, and roughly how soon that might be.

Why do you need more than one kind of data?

Because each kind sees a different group of members at a different moment. Claims see everyone who has care, but this is backwards-looking. And some members have no claims at all when they’re starting a journey. App activity sees a member early, sometimes the day the need shows up, but only for the members who come to us first. Context like plan design applies to everyone, but it does not give specific information, like intention or decision timing.

Put them together and you cover far more members, far earlier, than any one of them could alone. The point is breadth. Most members show up in just one of those sources, so if you only listen to one, you miss everyone else.

What’s different about the data HealthJoy has access to?

What’s different is where HealthJoy sits. We’re the front door for the whole benefits experience, so members come to us with questions long before there’s a claim: at enrollment, when they’re choosing a plan, when they get a bill they don’t understand, when they’re looking for a doctor. That gives us a record of how members actually behave. It is constantly evolving, and sits alongside the claims and eligibility feeds we get from the plan.

Most tools in this space have one or the other. The combination is what lets us see a need early and act on it in the same place the member already uses and trusts.

What’s the biggest misconception about “predictive” in benefits?

That it means certainty, like a crystal ball that says this person is having surgery on the fourteenth. That’s not what it is and it shouldn’t be. What we’re doing is identifying members who are more likely to be heading toward a certain type of care, early enough to be useful, so we can offer guidance. 

The second misconception is that a prediction on its own is worth something. It only matters if it changes what happens next: a message, a conversation, a better option surfaced while the member is still choosing. It is this linkage to engagement and real-world usage that makes the whole system work.

Why focus on surgery, imaging, and infusion?

Because they’re where a small number of decisions drive a large share of spend and because they’re plannable. There’s time between the need and the care, which means there’s time to help.

They’re also the services where the price depends enormously on the setting: the same MRI at a hospital outpatient department versus a freestanding imaging center, the same joint replacement at one facility versus another, the same infusion in a hospital versus an office or home. The clinical service is the same, but the cost is not. Imaging and surgery are where we’re furthest along. Infusion is a smaller population and a different kind of decision, so we’ve been more deliberate there.

What it means for HR leaders

How do you think about doing this in a way that helps members and doesn’t feel intrusive?

Our goal is to be helpful at a time when someone has a lot of questions and is trying to balance a number of needs like care, cost, benefit structure, or availability. We offer guidance about the care that’s in front of them: here’s how to find a high-quality, a lower-cost option, here’s what your plan covers, and here’s a person who can help.

What should an HR leader or consultant ask about timing when they evaluate a navigation vendor?

I’d ask four things. First, what is the earliest signal you look at? Is your trigger the claim or something earlier? If the answer is the claim, the member has usually already chosen.

Second, how often does your picture refresh? Some tools re-score once a quarter or once a year, and by then the window has come and gone. Within a day is the standard I’d look for.

Third, what do you do when you know? Identifying a member early and putting them in a report is not the same as reaching them with guidance while the choice is still open.

And fourth, what happens for members who haven’t had a claim yet? Those are often the exact people you most want to reach and a claims-only tool can’t see them.

If an HR leader remembers one thing from this conversation, what should it be?

The cost of a care episode is mostly decided in a short window before the first claim appears. Anything that reaches your members earlier in that window is where the opportunity is found. That’s what we’re building toward.

Reactive navigation answers the questions members ask. David’s work starts earlier, in the window where a need is visible but the choice isn’t made, and where guidance still has room to matter.

© 2026 HealthJoy. All rights reserved.

© 2026 HealthJoy. All rights reserved.

© 2026 HealthJoy. All rights reserved.