38.28%
Painting payroll plus benefits
Compared with 18.98% for materials and parts.
We analyzed 1,192,695 anonymized 2024 bill records. In the midpoint record set, 69.5% of bills were below $500 but represented only 7.9% of the billed amount.
The operating cost benchmarks include accounts payable benchmarks, a vendor concentration benchmark, contractor vendor spend patterns, monthly volatility, and a test showing why software audit false positives matter.
The CEO agenda is clear: automate the high-volume, low-dollar work, negotiate the five-vendor core that held 72.5% of billed amount, and manage one-time vendors as a control problem rather than the main savings pool.
1.19M
anonymized 2024 bill records
652
company IDs assigned by source systems
72.5%
billed amount in the five largest vendor names
2.2x
peak month versus a typical active month
Important context: the source systems assigned 652 company IDs to these records. We did not independently verify that every ID is a unique contractor. A bill is an amount owed, not proof of cash paid, material used, overpayment, or savings.
Public industry baseline
Level calculated each cost as a share of aggregate industry receipts using the U.S. Census Bureau's 2022 Economic Census. Across the selected trades, materials and parts ranged from 18.98% for painting contractors to 40.33% for glass contractors. That 21.35-point spread is why one universal contractor cost target is misleading.
38.28%
Painting payroll plus benefits
Compared with 18.98% for materials and parts.
40.33%
Glass materials and parts
The highest selected material share, against 30.23% for payroll plus benefits.
4.7%
Site-prep gasoline and diesel
About 2.19 times the aggregate specialty-trade share.
Current labor context
BLS split the average into $35.54 of wages and $15.69 of benefits. Benefits therefore added about 44 cents per wage dollar before vehicles, tools, supervision, nonproductive time, or overhead.
Average employer cost per employee hour worked for the broad construction industry, not a trade, role, wage quote, or fully burdened billable-hour cost.
NAICS 238150, 6,618 employer firms
NAICS 238160, 24,082 employer firms
NAICS 238210, 78,975 employer firms
NAICS 238220, 112,088 employer firms
NAICS 238320, 38,036 employer firms
NAICS 238910, 39,460 employer firms
How to read this: These are ratios of aggregate 2022 industry dollars for employer firms. They are not median-company percentages, SMB-only benchmarks, Level customer results, or recommended budgets.
The annual workload
The midpoint company record had 681 bills and 48 vendor names in 2024. That is too much for line-by-line management, but the dollars were concentrated enough to split the work intelligently.
Review terms, repeat baskets, freight, availability, rebates, credit, and service levels here.
Use separate rules for routine, emergency, specialty, and one-time vendors.
One-time vendors were 33.3% of names, creating approval, coding, tax-document, and fraud-review work.
Only 3.8% of billed amount sat in the one-time tail at the midpoint. It is not the first negotiation target.
Common vendor patterns
This table shows how many reviewed company IDs had at least one bill from each common vendor. It is not market share, contractor adoption, cash paid, or proof that one supplier is better.
370 company IDs with a bill
361 company IDs with a bill
305 company IDs with a bill
295 company IDs with a bill
263 company IDs with a bill
261 company IDs with a bill
243 company IDs with a bill
206 company IDs with a bill
197 company IDs with a bill
195 company IDs with a bill
*The recorded currency is assumed to be USD but was not independently verified. Billed amount is not cash paid, expense, material use, or savings. Typical means the midpoint among records using that vendor.
Home Depot appeared in 51.3% of records when we required at least 10 valid bills, compared with 61.1% when we required at least 50. Grainger appeared in 46.7% of records when we required at least 10 valid bills, compared with 63.7% when we required at least 50. The lesson is simple: vendor percentages are useful only when the minimum activity rule is stated.
Discount reality check
Home Depot
12.2%
of 370 records using this vendor had at least one bill at or above the $2,500 comparison threshold.
Lowe's
4.1%
of 195 records using this vendor had at least one bill at or above the $2,000 comparison threshold.
Amazon
0%
of 197 records using this vendor had at least one bill at or above the $10,000 comparison threshold.
For one AGC and Home Depot program example, 74 of 370 records using Home Depot, or 20%, cleared the $12,500 half-year threshold at least once. Applying 2% to the full $6.03 million pool gives $120,642. Because enrollment, registered payment, exclusions, returns, and actual rebate receipt were not visible, this is the maximum worth investigating, not achievable savings.
Software review accuracy
One read-only Level review initially flagged 206 recurring items and $10.68 million. After checking accounts, vendors, and business purpose, 42 software-related items totaling $197,066 remained for human review.
79.6%
of initially flagged items were removed after basic classification.
98.2%
of initially flagged dollars were removed. The remaining amount is still not waste or savings.
This is one Level review, not a market benchmark. The practical lesson is to classify recurring activity before asking owners to defend software.
Monthly swings
2.2x
midpoint peak month compared with a typical active month
A spike can come from seasonality, a project start, equipment purchase, accounting timing, or a data problem. It does not automatically mean costs rose or the peak was avoidable. Compare each business with its own recent pattern.
The midpoint record had 18.6 active vendor names in a typical month. That makes a monthly exception review more practical than treating every annual vendor relationship equally.
How we did the analysis
1. Clean the records
Keep the latest 2024 version of each named-vendor bill, remove deleted records, and exclude nonpositive or unusually large entries.
2. Measure each source separately
Calculate volume, vendor mix, and monthly patterns within each company ID so one large source cannot dominate the result.
3. Publish privacy-safe results
Report midpoint figures and only common vendor names with large groups. Keep identities, individual transactions, and small vendor groups private.
What we deliberately leave out
Customer identities, individual bills, uncommon merchants, small software-vendor groups, and any claim that billed amounts equal cash paid, overpayment, or realized savings. Technical audit references remain in the downloadable research record for analysts who need them.
Not yet. It is Level research across 652 company IDs assigned by the source systems. We did not independently verify that every ID represents a unique contractor, so the figures are management reference points rather than market percentiles.
No. A bill records an amount owed. It does not prove payment, material use, job cost, negotiated price, overpayment, or savings.
The ten common vendor families have enough records to show how big-box, trade-supply, and rental bill patterns differ. We do not publish uncommon merchants or small software-vendor groups.
Yes. The downloadable JSON includes the definitions, sample rules, calculations, and audit references. The CSV exposes the published findings and vendor rows. Raw customer records remain private.
Drop your info and we’ll show your real margin after callbacks and rework, which crews and jobs are worth the time, and how much cash is sitting in work you have already finished.
No commitment. Real numbers, not generic advice.