When we began launching a post-discharge support program, it became clear almost immediately that we couldn’t design it well from a distance.

This was our first time building a program like this with a large health system. We piloted it at a single hospital with a deliberately lightweight design, optimizing for speed and learning. Our initial referral model made sense on paper: a case manager identifies an eligible patient, explains the program, and submits the referral.

Simple enough, right? Turns out, it wasn’t. We weren’t seeing the referral volume or conversion we expected… signaling a deeper issue in the flow itself.

And that’s when it hit me: to design something that actually worked inside a hospital, we needed to be there. Observing daily workflows. Seeing real constraints. Understanding what clinicians, patients, and caregivers were navigating in practice.

So I went onsite with Starlight’s CEO and Chief of Healthcare Partnerships to do field research. We wanted to see how discharge actually happens, talk with patients, caregivers, and staff in context, and understand where our assumptions held up and where they broke.

This let us ask questions we couldn’t realistically answer from a distance:

• Where exactly was the process breaking down?
• What did eligibility identification and referral submission actually look like in case managers’ day-to-day workflows (i.e., time, friction, and cognitive load)?
• How was the program introduced to patients, how was value communicated, and how did patients actually react in that moment?
• What state of mind were patients in while hospitalized and at discharge, what messages resonated, and where did gaps in care show up?

Insight #1: The original referral model missed the majority of patients who actually needed support

On our first day onsite, it became clear the problem was baked into the model.

Case managers were responsible for introducing patients to Starlight and submitting the referral. The problem was that patients assigned a case manager were, by definition, already surrounded by support. Case managers were focused on the most complex cases… roughly the top 20%.

As I interviewed patients throughout their hospital stay, I started to see the support landscape around them more clearly. Many were being discharged to skilled nursing facilities. Others had home health services in place or deeply involved caregivers. One patient I spoke with had 8 different family members rotating throughout the day to make sure he was never alone.

These patients were complex, high risk, and absolutely deserving of attention.
But they were not unsupported.

Meanwhile, the remaining 80% of patients weren’t interacting with a case manager at all. They were still going home with medication changes, follow-up appointments, and discharge instructions they might not fully understand. Once they left the hospital, there was often no one proactively checking in.

Some of these patients didn’t need intensive, ongoing care. Most would recover just fine. But when questions came up (and they often did) there was no clear place to get real-time guidance on: which symptoms needed attention, how to handle medication changes, or how to stay on track with follow-up appointments.

Small gaps like these are often what turn manageable recoveries into avoidable readmissions. That gap is where the opportunity really was.

Insight #2: Patients are asked to learn when they’re least able to

Spending time on the floor made one thing clear: discharge is chaotic.

One patient I interviewed was interrupted constantly:
A doctor came in.
Then a nurse.
Then a case manager.
Then someone changing the bedpan.

There was very little uninterrupted mental space to process anything new.

Discharge nurses were well aware of this reality. They told us they do their best to emphasize the most important parts of discharge paperwork, but they also know much of it doesn’t stick.

Patients are being asked to absorb critical information at the exact moment they are least able to learn. Not because anyone is doing a poor job, but because the system optimizes for completion, not retention.

As we spent more time onsite, we discovered the discharge lounge, where patients across all levels of complexity waited after being formally discharged and before being picked up. The discharge lounge manager shared that she is often asked to walk patients back through their paperwork, especially medication changes layered onto already complex regimens.

When we talked about the idea of referring patients directly from the lounge, she lit up. She immediately saw the value in connecting patients to Starlight in that moment and walking them through their paperwork while questions were actually coming up.

The discharge lounge wasn’t just a waiting area. It was a moment of readiness. Patients were calmer. They were asking questions. They were actively trying to understand what came next… and we decided to treat that moment as an opportunity.

Insight #3: Caregivers didn’t need support… until they did.

Before going onsite, we knew caregivers played a decisive role. In early outreach, they were often the ones answering the phone, influencing decisions, and sometimes declining support on the patient’s behalf.

What we didn’t fully understand was why.

In the hospital, many caregivers I interviewed felt like they already had it covered. They were already planning on managing the discharge paperwork, tracking medications, and coordinating logistics. So when the program was introduced, it didn’t feel necessary yet.

But when I asked what happened the last time the patient was discharged, caregivers described running into questions once they got home… medication side effects, new symptoms, uncertainty about what was normal versus concerning. And when those questions arose, there was no clear place to turn for real-time guidance.

The issue wasn’t lack of need… it was timing. In the hospital, caregivers felt confident. After discharge, they were often left making clinical judgment calls alone. We needed to show up in that gap.

Insight #4: Post-discharge support isn’t always clinical

Spending time onsite made it clear that illness doesn’t just create clinical needs. It creates emotional strain that ripples through patients and their families.

I watched family members process diagnoses in different ways. Sometimes they disagreed about what information should be shared (and with whom), or how decisions should be made. That tension often turned into conflict, creating real emotional strain for everyone involved.

A few patients I spoke with had sick spouses at home they were expected to care for. Others were managing their own diagnosis while feeling guilt about how it was affecting their family. In those moments, it wasn’t about understanding discharge instructions. It was about emotional support, and our program needed to reflect that.

How field research changed the service design

We used what we learned onsite to reshape both the program and the product.

• We treated timing as a core product decision, designing for the reality that patients and caregivers move in and out of readiness. We built the ability to assess existing support, intentionally snooze patients who were already covered, and re-engage later when questions surfaced and our support would actually be additive.

• We stopped optimizing solely for case manager referrals and instead partnered with hospital staff to identify new referral moments and in-hospital roles that could reach patients the original model never touched, including fallback paths when referrals didn’t happen upstream.

• We redesigned how and when the program was introduced inside the hospital. Rather than relying on verbal explanations in chaotic moments, we created QR-based flyers with short videos that patients or caregivers could engage with asynchronously. This improved clarity, reduced cognitive load, and made it easier to revisit the program later.

• We added a 24-hour nurse line in direct response to caregiver needs, giving patients and families a real human to call when something felt off, instead of forcing them to make clinical judgment calls alone late at night.

• We began exploring our definition of post-discharge support and whether we should consider emotional and family-level needs. Based on what we observed onsite, we’re exploring bringing licensed clinical social workers into the program to support patients and families through emotional strain, family dynamics, and complex decisions.

Onsite field research as a go-to-market requirement

Seeing the hospital in person helped us understand the experience from multiple angles at once: what discharge feels like for patients, what decision-making looks like for caregivers, and what constraints clinicians are operating under minute to minute. This revealed new opportunities inside the hospital itself, new moments and touch points where support could be introduced more naturally and with far less friction.

Just as importantly, being onsite built familiarity. By the time we expanded across the hospital, staff already knew who we were and what we did. That recognition made change management and onboarding significantly easier.

Field research isn’t optional when you’re designing care that has to work both virtually and inside a hospital.