M7 Health

Your Scheduling Data Already Knows Who's About to Leave

Most conversations about nurse turnover start at the exit interview. What we've learned working with nurses, managers, and leaders across health systems of every size is that the warning signs show up long before a nurse leaves. And they show up in the scheduling data that health systems already have access to.

The problem is that nobody's connecting the dots

Here's why hospitals miss it. The data exists, but it lives in different places. Callout logs are in one system, nurses write down their availability on paper, float assignments are tracked on a whiteboard or buried in a spreadsheet. Nobody steps back and looks at a single nurse's pattern across all of it over time.

And because no one is connecting it, a callout looks like a callout. A preference submission looks routine. A float assignment looks like a staffing decision. Each data point, in isolation, means almost nothing. Strung together over months, they can effectively identify who’s thinking about leaving.

That's the shift in how we think about this. We're not measuring against a policy threshold. We're measuring against a person's own baseline and their unit's norm. By looking at changes and trends, we can draw actionable insights for leaders.

What we actually look for

We've spent years sitting with nurses, managers, and CNOs across health systems of every size, and we've looked at a lot of signals. Some vary by system, by unit culture, by how schedules are built. But a handful show up consistently, everywhere, as reliable early indicators. These are the ones that always seem to play a role.

Callout frequency. The most visible signal, but only meaningful in context. When callout frequency climbs against someone's own historical baseline, and especially when it diverges from what's happening across the rest of the unit, that's worth paying attention to.

Preference behavior. This one has two layers. Repeatedly overriding someone's work preferences builds frustration over time, and that frustration accumulates quietly. But the stronger signal is when a nurse stops submitting preferences entirely. They've stopped trying to shape their schedule. Disengaging from the scheduling process tends to precede disengaging from the job.

Float burden. Floating isn't inherently the problem. Inequity is. When one nurse is floating significantly more than her peers, that imbalance tends to show up on resignation lists quarters later. What makes this particularly worth addressing is that it's often one of the easiest things to rebalance once it's visible. The fix usually isn't complicated. Getting the visibility is the hard part.

Open shift pickup. When someone who used to grab extra shifts stops while their peers' rates hold steady, they're pulling back from the unit. It's a quiet signal, but a consistent one.

Overtime and workload drift. This is the one that surprises people most. Sustained overtime looks like commitment. It often signals exhaustion. The nurse who always says yes, who never pushes back, who covers every hole, needs a check-in just as much as the one who's started saying no.

Knowing who's at risk is only half the equation

Even a manager with good instincts faces a real structural problem. According to McKinsey's 2025 Nursing Pulse Survey, nurse managers average around 30 direct reports but may oversee as many as 250. They're managing daily staffing, callouts, float coverage, and everything else that comes with running a unit. A gut feeling about one nurse competes with a hundred other things demanding attention that shift.

That's where prioritization matters. The difference between a check-in that happens and one that doesn't often comes down to whether a manager was told, clearly and at the right moment, that this person needs a conversation today. Not a report to dig through. Not a dashboard to remember to open. A clear prompt, with enough context to act.

And acting doesn't have to mean a formal sit-down. Sometimes it's a text acknowledging a work anniversary. Sometimes it's flagging someone for a schedule adjustment before the next cycle goes out. Sometimes it's a five-minute conversation at the start of a shift.

This works at the unit level too

Individual signals matter, but so does how they cluster. When multiple nurses on the same unit start showing at-risk patterns around the same time, the issue usually isn't the individual nurses. It's unit conditions: the staffing model, how float assignments are distributed, the shape of the schedule itself.

Seeing the pattern at the unit level tells you where to look for the systemic fix. It shifts the question from "who do I need to check in with" to "what is this unit's environment doing to people."

A note on pulse surveys

We still use them and believe in them. Surveys tell you what people are thinking. Scheduling data tells you what they're doing. Both matter. The difference is that behavioral signals update continuously. You're not waiting ninety days for the next survey to find out someone is disengaging. The data is generating itself every shift, in systems you already have.

The bottom line

Every hospital we talk to is already generating this data. Most aren't using it. Even when leaders have a sense that someone is struggling, there's no clear mechanism to surface it, prioritize it, and make it actionable.

The vendors who make the biggest difference here aren't just the ones who can aggregate all of that into one place. They're the ones who can feed it to leaders at the right time, at the right level of detail, with a clear path to action. Not a monthly report. The right signal, to the right manager, early enough to do something about it.

You don't have to accept finding out at the exit interview. The data is telling you earlier. The question is whether your systems are built to help you hear it.

If you're thinking about how to build that capability at your organization, we'd love to talk through what we've seen work.