Nurse scheduling AI: Mercy's $30.7M and three other measured results

Will Drewes

Founder, Fern Strategy · LinkedIn · 2 min read

Four hospitals that measured it

Nurse scheduling looks like an administrative cost until you price what bad scheduling does: travel nurse premiums, turnover, charge nurses burning 30 hours a month on spreadsheets, and staff who think the schedule is rigged against them. The numbers below come from hospitals that measured before and after.

Mercy: $30.7 million in 2023. The savings came from lower travel nurse and premium labor spend after Mercy deployed AI-assisted nursing workforce management. Fill rate moved from 83% to 86%. Turnover dropped 8%. Time spent on staffing and scheduling tasks fell 20%. Small numbers on their own. Compounded across a health system they produced $30.7 million in a single year, and the lever was substitution: fewer agency fills, more internal coverage, less scrambling.

Mott Children's Hospital: 22 hours to 4. Pediatric on-call scheduling at Mott had been running 22 to 28 hours of planning time per month. After automating that workflow, the same work took 4 to 6 hours. For a charge physician running this on top of clinical duties, that's roughly three workdays back every month.

A JMIR pragmatic study: 81% less scheduling time, 74% fewer errors. A pre-post study in JMIR Nursing tracked one hospital's implementation. Monthly nurse scheduling time dropped from 32.0 hours to 6.0 hours, an 81.2% reduction. Scheduling error rate fell 73.8%. Adoption reached 94.9% of nurses (148 out of 156), which is the number most implementations quietly miss. The coefficient of variation in shift distribution fell 50%, so shifts were spread more evenly across staff. That metric rarely shows up in a vendor deck, and it is tied directly to retention and grievance rates.

An ANA case study: 12 hours to 2 minutes. A hospital profiled by the American Nurses Association reported that schedule builds and adjustments that used to take over 12 hours were done in 2 minutes. Staff satisfaction improved 56%. Perceived bias in scheduling dropped 30%. Labor cost savings ran about $300 per week. Scheduling fairness is one of the most persistent complaints in nursing and almost never gets a number put on it, so the 30% is the figure I'd pause on.

What this means if you run a health system

These results span different system sizes, specialties and study designs, so read them as direction. The pattern holds, though. The returns come from administrative time, which you get back immediately and can measure, and from premium labor substitution, which takes longer and compounds.

The equity and satisfaction numbers are harder to move and harder to measure, which is why they get cut from business cases. Turnover in nursing runs $40,000 to $60,000 per departure by most estimates. A 30% drop in perceived scheduling bias belongs in the model.

The governance question comes before build versus buy. Scheduling touches union contracts, labor law and accreditation. Whatever system you run needs a complete audit trail of who approved what and when. Get that right before you tune anything.