Use per diem for shift-level gaps and call-outs, travel or local contract for a defined block of eight to twenty-six weeks, and permanent when the seat is structural. The expensive error is running contract labor for eighteen months on a role that was always permanent — that costs a facility far more than the agency fee it was avoiding.
Use it for: call-outs, ratio protection, predictable weekly gaps, and building flexibility without expanding core FTE.
How it bills: hourly, shift by shift, no minimum commitment.
The catch: you are competing for the same per diem pool as every facility in your market. Block booking and recurring commitments get you priority; ad hoc requests do not.
Use it for: leave coverage, seasonal surge, EMR go-lives, service line expansion, and bridging to an approved FTE.
How it bills: hourly bill rate over a defined term, typically 8–26 weeks, with extension options.
The catch: the rate is set at booking, so short-notice starts price higher. Predictable needs booked early cost meaningfully less than the same coverage arranged with two weeks’ notice.
Use it for: structural roles, leadership, and anything you expect to still need in two years.
How it bills: contingent — a percentage of first-year base salary, invoiced when the hire starts, with a replacement guarantee.
The catch: it takes longer, and hiring permanently while critically short-staffed is how facilities end up with a bad twelve-month hire.
Work with the clinician on assignment first and convert at term with no additional fee. It is the lowest-risk route for hard-to-fill core roles, because competence on your own units, equipment and protocols is obvious within a few weeks in a way it never is in an interview.
It works particularly well in imaging and respiratory, where modality and vent-population competence are difficult to assess from a resume.
Contract to stabilize now, permanent search running in parallel without pressure. The traveler holds the schedule so the permanent hire is made on merit rather than desperation.
Cost comparison across all three models is published here.
Describe the gap and we will tell you honestly which model fits — including when the answer is neither.