The Contractor Turnover Treadmill: Construction Hired 3.99M and Lost 3.97M in One Year

Almost 4 Million Hires Against Almost 4 Million Departures

Here is the treadmill in one line. In calendar 2025, US construction made 3,987,000 hires and recorded 3,966,000 total separations. The industry averaged 8,267,000 jobs that year. The difference between those annual flows was about 21,000 events, roughly a quarter of one percent of average employment. The flows do not identify replacement hires, individual worker retention or the change between two employment-stock observations.

Read it as a rate and it is starker. Monthly hires averaged 4.0% of employment. Monthly total separations also averaged 4.0%. The 3,966,000 annual separation events divided by 8,267,000 average jobs is about 48%. That is an event-to-average-employment ratio: the same worker or job can contribute more than one event. It is not the probability that one employee leaves, and it is not a ceiling for any occupation.

Source: BLS Job Openings and Labor Turnover Survey, construction sector, annual totals, not seasonally adjusted. Employment base from BLS Current Employment Statistics, construction, 2025 annual average.

Note which line is bigger. Layoffs and discharges (2.06M) exceed quits (1.74M). That is the shape of a project-driven industry: crews scale up for work and scale down when it finishes. Some of that churn is structural and you would not want to eliminate it. What you want to eliminate is the expensive, avoidable kind, and the way you find it is by measuring your own separations rather than assuming the industry number is yours.

This is the treadmill. The financial cost can be larger than the recruiting line suggests, because it also includes the billed-revenue gap when a journeyman seat (Level rate-card median $76/hr) is covered by an apprentice (rate-card median $39/hr) for weeks while you backfill. That rate-card field is entered inconsistently across companies, so treat the $37 spread as an illustrative input, not a certified price gap.

---

Why There Is No Trustworthy "Turnover Rate by Role"

You will find per-role construction turnover tables all over the internet, running from the high eighties for general laborers down to the low forties for administrative staff. We published one on this page for over a year. We have removed it, and here is why.

BLS JOLTS is the national industry-flow series used here, and it reports these construction flows by industry, not by occupation. There is a construction series. There is no electrician series, no project-manager series, no administrative-staff series. So a per-role ladder cannot be sourced to JOLTS, no matter how often it is attributed there. When we went looking for the origin of the ladder we had published, the citations behind it could not be located and the numbers appeared in no dataset we hold.

There is a second problem: an industry average is not a role-level ceiling or target. Construction total separation events were about 48% of average employment in 2025, but that does not identify an electrician, foreman or office-manager rate. Different subgroups can sit above or below the average. We cannot infer an occupational ranking or retention target from these aggregate flows.

We also cannot derive it ourselves. Our contractor dataset has employee records, but it does not carry termination dates, so it cannot produce a turnover rate for any role. Publishing one anyway would be inventing it.

What to do instead. Count your own separations by role over the last 12 months. Divide by your average headcount in that role. That is your turnover rate, it is specific to you, and it is the only version of this number that can tell you where to spend retention money.

---

What Each Departure Actually Costs

Most contractors think of turnover cost as the recruiting expense. It's not even close. The real cost includes recruiting, onboarding, training, lost productivity during ramp, and the billing rate gap while the seat is filled with a less-skilled worker. Billable revenue scales with skill, but the dollar amount depends on your rates and hours. Level rate-card data across 18,000+ tracked field employees shows median rates of $76/hr for journeymen, $39/hr for apprentices and $34/hr for helpers. That field is used inconsistently (some companies enter customer billing rates, others loaded cost or base wage), so neither the level nor the spread establishes pricing until each field definition is matched (see our bill rates by skill level). As fictional arithmetic only: 1,900 billable hours x $76 is about $144,400 a year for a journeyman, and 1,600 billable hours x $34 is about $54,400 for a helper. Those hour counts are assumptions, so swap in your own billed hours and rates. The replacement costs in the table below are also illustrative planning assumptions. We have not located a dataset, sample size or year behind them, so treat every row as an editable input, not a measured median. The $4,500 to $12,000+ per skilled departure range that circulates in industry articles is likewise unattributed here.

In this illustration, the $12,800 input covers recruiting, onboarding and incremental training spending only. It excludes lost billed revenue and the productivity gap, which are modeled separately. If your replacement-cost estimate already includes those losses, remove them before adding the gap again. A skilled tradesperson therefore costs an assumed $12,800 in this spending bucket, before the assumed 8-12 weeks of reduced output while they ramp. The illustrative project manager row assumes $45,300 and up to 20 weeks to full productivity, which would mean about five months of a PM operating below capacity on your most complex jobs. Replace each row with your own recruiting invoices, onboarding hours and observed ramp time.

As fictional arithmetic, a $30/hr tradesperson can cost $55-$60/hr fully loaded, about 1.8 to 2.0 times base pay, once you factor in payroll taxes, workers comp, benefits, truck, tools, and training. Your own payroll and fleet costs set the real multiplier. The cost of hiring a new tech runs through the full build. At the illustrative $55/hr loaded rate, 2,080 paid hours comes to about $114,400 a year.

---

The Math on a 20-Person Crew

Let me make this tangible with a worked example rather than a borrowed rate. Run it with your own separation count, not ours.

Take a fictional 20-person field crew. Total separations include project-end layoffs, which usually do not create a seat to backfill, so this example uses quits instead. Construction quits were 1,741,000 in 2025 against an average base of 8,267,000 jobs (BLS JOLTS and CES), about 21.1% a year. That gives roughly 4 quits across a year (20 x 0.211 = 4.2). This is an industry-average stand-in, and your real number could be well above or below it, which is the entire reason to count yours. Count only the departures you actually had to backfill.

Direct replacement costs: 4 quits x $12,800 (illustrative) = $51,200

Billing rate reduction during backfill: When a journeyman leaves and you temporarily fill the slot with an apprentice, you lose billed revenue at the rate spread. From our bill rate benchmarks by skill level, the rate-card median is $76/hr for journeymen and $39/hr for apprentices (a field entered inconsistently, so neither its level nor spread certifies customer pricing). If each quit creates a fictional 6-week gap at the lower rate, with 40 billable hours per week as an explicit fictional assumption:

4 x 6 weeks x 40 hours x ($76 - $39) = $35,520 in lost billed revenue. This is revenue, not margin. The apprentice is also paid less than the journeyman, so the margin loss is smaller than $35,520 unless the seat sits empty.

Replacement spending plus gross billing gap in this fictional scenario: about $86,720

That is roughly $87,000 of combined replacement spending and gross billing exposure in the fictional crew. It is not an expense total, cash loss or net retention benefit: reconcile wage offsets, actual billable hours and overlap before making a budget decision. Counterexample: if you treated all 10 expected total separations as backfills, including project-end layoffs, the same inputs would give 10 x $12,800 + 10 x 6 x 40 x $37 = $216,800 (about $217,000). That overstates replacement cost for most crews, because a layoff at the end of a project usually does not create a seat to fill. Swap in your own count of backfilled departures and the arithmetic is the same.

For context, the Level Index company-level average billed labor rate has a median of $79/hr across 1,770 rate-metric-eligible contractors within the broader 2,242-company research universe, with the same caveat that the rate-card field is entered inconsistently (more in what we learned reviewing contractor financials). We do not have a verified revenue-per-technician distribution to publish here. The utilization benchmarks can compound the problem. Utilization is billable hours divided by total paid hours, and Level's working range is 65-80%. As a fictional example, a new tech at 60% during ramp versus a veteran at 75% loses volume as well as rate. Do not confuse utilization with the 97.1% median billing capture rate, which divides invoiced hours by hours logged on jobs.

---

The Industry Shortage Makes It Worse

This would be painful in a normal labor market. In this one, it's devastating.

ABC's January 15, 2026 forecast estimates 349,000 net new workers needed in 2026 and 456,000 in 2027. These are modeled annual demand figures, not current unfilled positions. An earlier ABC forecast put 2026 demand at 499,000, so do not combine the two forecast vintages into a 2026-2027 range. In the August 28, 2025 AGC/NCCER survey, 92% reported hiring difficulty and 45% said shortages delayed projects. Those describe survey respondents, not every firm. For current openings, use JOLTS rather than substituting a demand forecast.

When your journeyman leaves for $2/hr more across town, the replacement timeline may not be 2-3 weeks. The table above assumes 8-12 weeks to productivity for a skilled tradesperson, and time to fill the seat comes before that. Track your own days from resignation to accepted offer. Every week that seat is empty is another week you're either turning down work, sending a less-qualified tech, or paying overtime to cover the gap.

Revenue per technician varies widely, and we do not publish a verified distribution or tenure effect for it here. Experienced techs who know your systems, your customers, and your dispatch patterns may produce more revenue per hour than someone in their first 90 days. To check, compare billed revenue per paid hour for techs in their first 90 days with techs past their first year, over the same months.

---

Where Retention Has Actual ROI

Using the illustrative inputs above, cutting 2 avoidable quits a year out of the fictional scenario preserves:

• 2 x $12,800 = $25,600 in direct replacement costs
• 2 x 6 weeks x 40 hours x $37 spread = $17,760 in billed revenue (gross, before the lower apprentice wage)
• $43,360 of replacement spending plus gross billing exposure in the illustration, before labor-cost offsets and retention spend

The levers that actually move retention aren't complicated, but they cost money, and the question is whether the investment pays back. A few to test against your own records:

Compensation bands may matter as much as total comp. Senior roles often churn less than junior ones, though we have not measured that in our own data, and the difference may not be just money. Clear progression can matter too. A clear progression path is a retention hypothesis to test with separation records and interviews. The $39 and $76 mixed rate-card medians are not pay bands or certified customer prices. Transparent pay progression tied to skill levels and certifications gives your workforce a reason to vest.

Test schedule predictability against avoidable quits. Our scheduling benchmarks cover overtime patterns. Chronic overtime and chaotic scheduling are plausible drivers of departures, but we cannot show from our dataset that contractors with tighter utilization ratios hold people longer, because it does not carry termination dates. Note that Level's utilization working range is 65-80% of paid hours, not 90-96%. A practical check: compare 12-month quits for techs with chronic overtime against techs without it.

Compare a raise with expected avoided replacement cost. A $3/hr raise for a skilled tradesperson costs $6,240/year (2,080 hours x $3, before payroll taxes on the raise). Replacing that same tradesperson costs an illustrative $12,800 plus $8,880 in billed revenue during the gap (6 x 40 x $37), or $21,680. If the raise has a 50% chance of retaining the employee for an additional year, the gross-impact screen is 0.5 x $21,680 = $10,840 before wage offsets, payroll taxes and contribution-margin effects; it does not yet prove positive net benefit. Counterexample: if fairness means giving the same raise to five techs to keep one, the cost is $31,200, and $31,200 divided by the gross $21,680 screen is about 1.44 quits. That is not a verified break-even threshold: use net avoided contribution loss and spending, not gross billing, for the decision.

---

The Honest Assessment

We do not have a measured turnover-cost range by company size. Using the illustrative inputs on this page, each avoidable skilled quit combines $12,800 of replacement spending with $8,880 of gross billing gap, a $21,680 exposure before labor-cost offsets ($12,800 replacement plus $8,880 of billed-revenue gap). Multiply that by your own count of avoidable quits. The illustration only becomes your cost estimate once actual spending and contribution losses are reconciled. The difference between contractors who manage it and those who don't is visibility, knowing the number, knowing which roles churn fastest, and knowing whether your retention spending actually produces ROI.

The treadmill only stops when you start measuring it.

---

Q: How does Level help with turnover cost analysis? A: We work from your payroll, field service software, and accounting data to estimate turnover cost by role, including the billing rate gap, productivity ramp, and recruiting spend, where those records carry separation dates and recruiting costs. The first audit is free, and the turnover analysis is part of it.

Q: What turnover rate should I target? A: Match the definition before comparing. Construction had about 4.0% monthly total separations in 2025; the annual event count divided by average employment is about 48%. Neither measure is a role-specific target, a worker-level departure probability or a ceiling. Count your own backfilled departures by role, distinguish quits from planned project-end layoffs, then measure the cost of avoidable departures. The goal is retaining productive capacity where its net benefit exceeds the cost of the intervention.

Q: Why did this post used to show turnover by role, and where did that table go? A: It showed a per-role ladder, from general laborers at the top down to administrative staff at the bottom, that we could not defend. The citations behind it could not be located, the numbers appear in no dataset we hold, and BLS JOLTS, the only authoritative source for construction turnover flows, reports by industry rather than by occupation. The skilled-trades rate it showed also sat above the total-separations rate for all of construction, a comparison we could not verify against any role-level source. We removed it rather than keep publishing it with a disclaimer. The industry flows above are what the data actually supports.

Q: How do I calculate my own turnover cost? A: Start with the number of departures in the last 12 months that you actually had to backfill (project-end layoffs usually do not count). Multiply by the replacement cost for each role (the $12,800 skilled-trades figure on this page is an illustrative assumption; use your own recruiting and onboarding costs if you have them). Then estimate the billing rate gap, how many weeks each seat was filled by a lower-skilled worker or left empty, and what that cost in revenue. Keep replacement spending and gross revenue gaps separate. Subtract avoided wages, use actual contribution on displaced work, exclude amounts already inside the replacement-cost input, and then add incremental overtime or documented disruption once. The resulting estimate is conditional on those inputs, not a guaranteed floor.

Q: Is it better to invest in retention or just hire faster? A: Usually retention, when the departures are avoidable quits. Hiring faster in a tight skilled-labor market means competing with every other contractor for the same people. Check the current JOLTS construction job openings series for the size of that competition. In the illustrative numbers above, a $6,240 annual raise costs less than half of one $12,800 replacement, though a raise given to several techs at once changes that math. Invest in keeping the people who already know your business.