You hire an assistant teacher in March. You train her, she settles in, the families like her, and by February she's gone. Most directors have stopped being surprised by this.
It's worth being surprised by it, because turnover isn't spread evenly. In the last national count, taken in 2019, 44% of American centers had nobody leave from the staff who work with under-fives. Another third lost more than a fifth of them.
That's a wide gap between centers doing the same work. Here's what the research says about who leaves, when, and which of the usual advice has evidence behind it.
The year that does the damage
Researchers followed 14,373 teachers at publicly funded childcare centers in Virginia for a year. Their paper hasn't been peer reviewed yet, so hold it lightly. Thirty-eight percent were no longer teaching at their center a year later.
Underneath it sits the finding worth your attention. Teachers with under a year in the field left at 54%. Teachers with five years or more left at 30%. And the decline isn't gradual. Most of it comes from getting someone through their first twelve months.
Assistant teachers left at 43%, lead teachers at 35%. So your risk sits with the newest, most junior person in the building, in her first year.
That should change where a director spends her attention. Plenty of centers pour energy into their veterans and leave the new assistant to find her own feet.
Nobody has tested a fix for this, so treat the next bit as opinion and not research. Give every new hire one named person who owns her first ninety days. Put ten minutes a week in the diary and keep it. And stop putting your newest assistant in the hardest room because that's the room that's short. That room is what finishes her.
The pay conversation, honestly
Almost every article on this says you can't compete with retail. Worth checking the number.
In the Bureau of Labor Statistics wage figures released in May 2026, covering 2025, childcare workers earn a median of $16.82 an hour. Retail salespersons earn $17.03. That's a gap of twenty-one cents. Against fast food work, childcare is $1.82 an hour ahead. I expected that number to be much worse, and I checked it twice.
Pay absolutely still matters, and the Virginia paper measures it. Teachers earning $12 to $14 an hour left at 45%. Teachers on $18 or more left at 27%.
But the gap that's actually pulling people away is the school district, and it's benefits. Berkeley's 2025 California study found 44% of center assistant teachers were on some form of public assistance. For transitional kindergarten teachers, teaching four-year-olds in public schools in the same state, it was 11%. Retirement savings ran at 45% for assistants and 84% for TK teachers.
They mostly do not go to another center
This part surprises owners. Of everyone who left a Virginia center, roughly four in five weren't found at any other center the state could see.
The Cleveland Fed found something starker nationally. About half of childcare workers who leave a job leave paid work entirely, against a quarter across all occupations.
So the person you lose is usually gone from childcare, not poached by the center across town. Matching a rival's wage by fifty cents aims at the wrong problem.
What the evidence supports, and what it does not
Here's the uncomfortable part. Most standard retention advice has never been tested.
One bit of housekeeping first. The turnover figure quoted in nearly every article on this topic, that 33% of American early educators leave each year, is a misreading. The federal report behind it says 33% of centers had high turnover, meaning they lost more than a fifth of their staff. Those are different claims, and the wrong one gets repeated until it reads like a fact.
Researchers reviewing 144 workforce strategies for a federal office reported no known studies evaluating any of the ones aimed at workload, staff support, or teacher wellbeing. Only five of the 144 were about teacher wellbeing in the first place. Recognition schemes, appreciation weeks, team-building, culture work. No evidence either way. That's a hard line to write the week after you've bought the pizza.
Two things that get recommended constantly have been measured, and neither showed a difference. Centers that paid for training had high turnover at 34%. Centers that paid for none ran at 31%. Centers whose directors had management training ran at 34%, against 30% for those whose directors hadn't. Neither gap was big enough to count.
One thing worked, and it was tested properly. In 2019 a Virginia county ran a lottery. Teachers at some centers were offered up to $1,500 to stay eight months. Teachers at the rest weren't. At the end, 15% of the childcare teachers who'd had the offer were gone, against 30% of those who hadn't. That's one county and about 600 educators, written up in a university brief instead of a journal. It's still the only retention idea in this field anyone has tested that way.
Benefits show up in the national data, though the evidence is mixed. Centers with no health insurance or retirement had high turnover at 40%, against 29% at centers offering both. A second federal study found the link faded once you allow for wages.
So if you have money to spend, the evidence points at pay and benefits rather than perks. If you don't, stop spending on things nobody has shown to work, and spend your own time on the assistant in her fourth month.
Seeing it before it happens
None of this helps if you can't see it. Most centers can't say how many staff are inside their first year, or which room lost three people last spring.
ChildcareWow keeps staff records, credentials and schedules in one place, and the scheduling is ratio-aware, so next week's rota stops eating a Friday afternoon. Those are the hours you'd otherwise not have for the person in month four.
It won't raise anyone's wage, and nothing in the research says software keeps teachers. What it buys you is time, and a clear view of who's new. That's what you need before you can act on any of it.
We built it because we run centers too.
Book a 20-minute walkthrough(https://booking.childcarewow.com/#/Demo). We'll show you your staffing and schedules the way your director would use them on a Monday.
SOURCES
• Amadon, Lin & Padilla, "Turnover in the Center-Based Child Care and Early Education Workforce," OPRE Report #2023-061, March 2023 — 2019 NSECE, 6,904 centers representing 121,049 nationally. 44% no turnover; 33% high turnover (defined as >20% of staff working with children leaving in 12 months). PD benefits 34% vs 31% (n.s.); director management training 34% vs 30% (n.s.); no health insurance/retirement 40% vs 29% (p<0.05) — https://acf.gov/sites/default/files/documents/opre/workforce_turnover_snapshot_apr2023.pdf
• Fares, Markowitz & Bassok, "Who Leaves? How Job and Teacher Characteristics Relate to Turnover in Child Care Settings," EdWorkingPaper 26-1417, Annenberg Institute, March 2026 — 14,373 lead and assistant teachers at 1,243 publicly funded Virginia centers, fall 2023 to fall 2024. 38% overall; <1yr 54% vs 5+yrs 30%; assistants 43% vs leads 35%; $12-14/hr 45% vs $18+/hr 27%. WORKING PAPER, NOT PEER REVIEWED — https://edworkingpapers.com/sites/default/files/ai26-1417.pdf
• U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics, May 2025 estimates, released 15 May 2026 — childcare workers (39-9011) median $16.82/hr; retail salespersons (41-2031) $17.03; fast food and counter workers (35-3023) $15.00 — https://www.bls.gov/news.release/ocwage.t01.htm
• CSCCE, UC Berkeley, "State of the Early Care and Education Workforce — California," 2025 California ECE Workforce Study, published Oct 2025, tables updated Jan 2026, Table 13 — center assistant teachers 44% on public assistance vs TK teachers 11%; any retirement savings 45% vs 84%. Spring 2025 fieldwork, weighted — https://cscce.berkeley.edu/wp-content/uploads/2025/10/state-of-ece-california-2025-updated-jan2026.pdf
• Bernardi, Hsueh & Roach (MDRC for OPRE), "Child Care and Early Education Workforce Recruitment and Retention," OPRE Report #2023-178, March 2024 — 144 active strategies reviewed. "There are no known evaluation or implementation studies of strategies that directly target workplace demands or supports, educator psychological well-being, or CCEE system alignment and inequities." Only 5 of the 144 targeted educator psychological well-being — https://acf.gov/sites/default/files/documents/opre/41672_HHS_OPRE_BASE_Brief_Environmental-Scan_v11_508.pdf
• Bassok, Doromal, Michie & Wong, "The Effects of Financial Incentives on Teacher Turnover in Early Childhood Settings: Experimental Evidence from Virginia," EdPolicyWorks / University of Virginia, December 2021 — lottery-based, Fairfax County, ~600 educators, 1 May to 31 Dec 2019. Child care subgroup (85% of sample): 15% turnover with the offer vs 30% without. Full sample 14% vs 25%. NOT PEER REVIEWED — https://researchconnections.org/childcare/resources/133036
• Fee, "Childcare and Education: Worker Flows," Federal Reserve Bank of Cleveland, 19 January 2024 — CPS microdata 2010-2022. ~50% of childcare workers leaving a job left the labour force, vs 24.1% across occupations — https://www.clevelandfed.org/publications/cd-reports/2024/20240119-childcare-and-education-workforce
