Caris Healthcare: Building Referral Strategy Around Care Impact

At a Glance

Client	Caris Healthcare, a hospice provider with 33 offices in five regions
Goal	Focus provider and patient education where it does the most good for patients and families
Approach	An office-level referral mix model built with regional directors and field leaders
Data	More than 40 percent of accounts had no facility type; categories were redefined and accounts reclassified
Adoption	Targets reviewed office by office with every regional director, plus a dashboard Caris maintains
Outcome	A projected 25 percent increase in average daily census, and data the field teams trust

The Situation

Caris provides hospice care through 33 offices in five regions. In each office, a small team of Caris Representatives works with hospitals, nursing facilities, physician practices and assisted living communities, helping clinicians, patients and families understand when hospice can help. Caris engaged me to find where that limited time would do the most good, set a target referral mix for each office, and build a dashboard to track progress.

Defining What Better Means

For most businesses, optimizing referral mix will mean optimizing around volume or revenue. But for Caris, the goal was framed around care. Erika Gennari, who led the project for Caris, stated, "Our goal is to do the most good for the most people." We built the model around a Care Impact Score that we developed, measuring each type of referral source by factors that impact care quality, including length of stay.

Length of stay matters because many patients reach hospice only in their final days, and they and their families miss much of what it offers, from symptom management to bereavement support. Sources that refer earlier give families months of support instead of days.

Focusing Education Where It Helps Most

No referral is turned away, and every one is still evaluated on the patient's needs. But the model guides where representatives spend their outreach time on provider and patient education, so more patients are referred while hospice can still do the most for them. Each office keeps its existing referral and admission targets, and the model recommends the mix of facility types that will get it there.

The Projected Impact

If each office focuses on the recommended mix, the model projects network-wide average daily census, the number of patients in Caris's care on a typical day, to rise by about 25 percent. Most of that gain comes from the change in mix, so it depends on where representatives focus their time. For families, a steadier census means higher-quality care, because offices can staff consistently and keep care teams with the same patients.

What the Data Revealed

Reaching that goal depends on data the teams can rely on. When the analysis began, more than 40 percent of CRM accounts had no facility type, others were misclassified or sat in a large "Other" category, and several hundred were duplicated or assigned to more than one office.

Reviewing these gaps with the regional directors showed the cause. Caris's 18 facility types overlapped ("long-term care" could mean assisted living or a nursing home), and staff entering data under time pressure guessed or left the field blank. We even found hospitals recorded as labs.

Those conversations changed the categories themselves. We merged overlapping types, reclassified accounts that could be identified reliably, and separated out self-referrals, which outreach can't influence. Caris is now consolidating its referral source definitions, which will shape how it captures data long after this model.

Cleaner data also makes the census goal easier to reach. Targets are set by facility type, so every misclassified or blank account weakens the guidance representatives rely on and the dashboard leadership uses to track progress.

Built With the Field

I worked directly with Caris's regional directors and field leaders throughout the project. Their input shaped the model's assumptions, including which sources teams could realistically influence and where local markets differed from the network average.

When the recommendations were ready, I walked through them with each regional director, office by office, so they understood how targets were set and could explain them to their teams. Where the numbers didn't match what directors knew from the field, we dug in, and more than once the field corrected the data.

What the Teams Valued Most

When I asked the regional teams what they were getting from the project, the answer I heard most often was the data cleanup. The model gave them targets, cleaner data gave them confidence that the numbers reflected their markets.

What Comes Next

Caris now has a referral mix model built on care impact and a clear view of the data foundation it needs. The next step is finalizing referral source definitions, assigning ownership of how account data is captured, and making sure future CRM changes support the teams' decisions. Erika and I plan to revisit the results in 2027.

On a personal note, this project mattered a great deal to me. I have seen firsthand how much difference good care at the end of life makes, for the patient and for the family around them. Working with Caris was a reminder that the decisions data informs are ultimately about people, and that getting the data right is part of getting the care right.