The float pool math nobody runs
Most systems know their agency spend and not their variance. Here is the calculation that tells you how much of your contingent labor should have been an internal hire — written by a staffing agency, which should tell you how confident we are in the answer.
Contingent labor is priced for uncertainty. You pay a premium because the agency carries the risk of finding someone quickly, for a short period, with no guarantee of future volume. That premium is fair when the need really is uncertain. It is money set on fire when the need is completely predictable and simply unstaffed.
The distinction is measurable, and almost nobody measures it.
Step one: separate baseline from peak
Take twelve to eighteen months of scheduling data for a single unit. For each week, record the number of shifts covered by contingent labor. Now sort those weekly numbers from lowest to highest and look at the bottom of the range.
That floor — the number of agency shifts you used even in your quietest weeks — is your baseline. It is not variance. It is a permanent staffing shortfall that you are covering at surge prices, every week, forever.
If your quietest week still used eighteen agency shifts on that unit, you do not have a staffing variance problem. You have roughly two unfilled FTEs and a purchasing habit.
Step two: price the two differently
For the baseline portion, compare the fully loaded cost of an internal float FTE — wage, differential, benefits, burden, and a realistic allowance for productivity and PTO — against the annual cost of covering those same shifts at your blended agency bill rate.
For most systems this comparison is not close. The internal hire wins clearly on baseline coverage. That is unsurprising: you are comparing a permanent employment relationship against a product priced for urgency and risk.
For the peak portion, run the same comparison and it usually flips. An internal FTE hired to cover a demand spike that occurs eight weeks a year spends forty-four weeks being expensive and idle, or being floated somewhere they are not competent to work — which is worse.
Step three: the part everyone gets wrong
Sizing the pool to the average. The average is the one number that describes neither state. A pool sized to average demand is simultaneously too small during peaks — so you still buy agency, at a rate that is now worse because your volume dropped — and too large during troughs, so you carry idle premium hours.
Size to the floor, not the mean. Cover the baseline internally, and buy the peaks. That combination is almost always cheaper than either extreme, and it is why “eliminate agency labor” is a bad goal. Agency has a legitimate job. It is just not baseline coverage.
The non-financial variables that decide it
The math tells you the pool should exist. Whether it works depends on things that do not appear in the spreadsheet:
- Differential. It has to be enough that a competent nurse chooses float work. Below your blended bill rate is a wide band — you have more room than you think.
- Notice. Float nurses tolerate uncertainty about where far better than uncertainty about when. Advance-notice rules retain more people than another dollar an hour.
- Competency clusters. Floating med-surg to med-surg works. Floating anyone anywhere destroys the pool within two quarters, because nobody wants to be perpetually incompetent at work.
- Scheduler behavior. If calling an agency is faster than releasing an internal open shift, your schedulers will call the agency — correctly, given their incentives. Fix the friction or the pool sits unused.
- Manager hoarding. Without governance, unit managers will treat pool staff as their own. That is rational and it defeats the entire purpose.
Where it does not work
High-acuity specialties with narrow competency requirements — ICU, OR, cath lab, L&D — often cannot float against anything. The pool has nowhere to send them on a quiet day, so the economics collapse back toward dedicated hiring or contract labor. Attempting it anyway is a common and expensive mistake, and it is usually visible within a quarter if anyone is watching.
Why we publish this
Because the alternative is worse for us. Systems that use agency for baseline coverage eventually notice, cap their spend, cut suppliers, and go to RFP angry. Systems with a healthy internal pool use agency for what it is genuinely good at — peaks, hard specialties, and speed — and they stay partners for years at a volume that is smaller but far more durable.