Janitorial And Security Services Employment Data By County Data: Dataset Structure and Field Coverage

The Quarterly Census of Employment and Wages (QCEW) — NAICS 56 record is the only source in Datadory's Diversified Support Services slice that publishes an industry-by-county quarterly panel. Fifteen fields define each row — Area Type, Area Name, Year, Time Period, Ownership, NAICS Level 0 through 6, NAICS Code, Industry Name, Establishments, Average Monthly Employment with first-, second- and third-month detail, Total Wages (All Workers) and Average Weekly Wages.

The two codes this page is about sit at the bottom of the NAICS hierarchy: 561720 Janitorial Services under 561 Administrative and Support Services, and 561612 Security Guards and Patrol Services beside it. Filter NAICS Code LIKE '56%' and set NAICS Level = '6' to reach them; Level 3 gives the whole Administrative and Support Services rollup for a county context line.

A county query needs no registration, no integration key and no payment:

Expect a lag of roughly five months between quarter close and publication — build pipelines around it rather than expecting a same-quarter read.

Suppression is the main analytical constraint on county-level work. BLS withholds any cell whose publication would disclose information about cooperating employers, so small county-and-six-digit-industry combinations — a rural county's handful of security-guard firms, say — can return no row even though covered employment exists there.

The practical consequence: a missing 561612 cell does not mean zero security employment, and summing published counties will undercount the state total. Check the suppression flag before computing shares or ranking counties by size, and fall back to the Level 3 rollup when a six-digit cell is blank.

Only the 2023–2025 Q4 resource was confirmed active in the CKAN datastore during verification; datastore-queryability of the older period files was never tested, so batch jobs against pre-2020 quarters should use the bulk CSVs.

Occupational Employment and Wage Statistics (OEWS) — support occupations prices the workers behind those establishment counts. It estimates employment and wages for roughly 830 occupations across about 580,000 rows / 112 MB in a single CSV spanning 2009 through 2026 Q1, free under Creative Commons Attribution 4.0.

Five SOC codes isolate the support-services labor pool: 372011 Janitors and Cleaners, Except Maids and Housekeeping Cleaners, 371011 First-Line Supervisors of Housekeeping and Janitorial Workers, 339032 Security Guards, 499071 Maintenance and Repair Workers, General, and 436014 Secretaries and Administrative Assistants. Each carries the 10th/25th/50th/75th/90th percentile wages plus employment counts by area.

Statistics of U.S. Businesses (SUSB) adds the enterprise-size cut QCEW lacks, tabulated from the Census Business Register down to county level with no registration. Its ENTRSIZE field classifies every firm by the employment of the whole owning enterprise — fewer than 5 employees, 5–9, and upward through 100,000+.

The national 2022 file shows why the split matters for janitorial markets specifically: NAICS 5617 Services to Buildings and Dwellings records 215,428 firms and about 2.1 million employees, and 149,120 of those firms (roughly 69 percent) have fewer than five employees — only 18,528 reach 10–19. Annual payroll sits near $81.9 billion against receipts of about $230.1 billion.

Read the pair this way: QCEW tells you how many janitorial workers a county employs; SUSB tells you whether they work for hundreds of owner-operators or a few regional chains — which changes who an RFP actually competes against. Cells flagged G/H/J/S carry injected noise and are approximations rather than exact counts, and receipts appear only in years ending in 2 and 7.

The honest limitation: neither source publishes an occupation-level janitor or guard count by local area the way OEWS does. Label cross-country figures with their actual grain — Canadian payroll employment by province and UK enterprises by SIC class are industry counts, not occupational ones.