Quarterly Services Survey (QSS) Data

It is the Census Bureau's only timely indicator of how America's service industries are performing, and this record is the Diversified Support Services shelf's copy of it. Each quarter the survey produces estimated total operating revenue, expenses and revenue share by class of customer for roughly 19,500 employer businesses in selected service sectors. Administrative and Support and Waste Management and Remediation Services - NAICS 56, the industry cluster behind outsourced work of every kind - has been measured since the first quarter of 2009.

The detail resolves usefully. Sector-level totals for 56 sit beside subsector cuts for 561 Administrative and Support Services and 562 Waste Management and Remediation Services, and inside 561 the industry lines name employment services (5613), travel arrangement and reservation services (5615) and other administrative and support services. The Bureau of Economic Analysis feeds the survey directly into GDP estimates, so these are not just statistics about the economy - they are part of how the economy gets counted.

Scale check: for the second quarter of 2026 the survey reported $360.8 billion in total revenue for NAICS 56, up 2.6 percent year over year, of which $317.8 billion came from administrative and support services. Benchmarking against the Service Annual Survey happens annually, keeping the quarterly path honest between censuses.

One row per industry line per quarter, exactly as a delivery lands:

Eight verified fields form the spine of a quarterly row - enough to identify the industry line, price it, track its movement and weigh the estimate:

The identity pair (2017 NAICS code, kind_of_business) pins each row to a named slice of NAICS 56, from the two-digit sector down through 5613 employment services and 5615 travel arrangement. total_revenue_millions carries the money - employer-firm revenue in millions of dollars for the reference quarter and its comparators, $360,752 million across NAICS 56 in 2Q2026. percent_change carries the movement in both directions worth having: quarter over quarter (up 5.9 percent for the sector) and year over year (up 2.6).

The statistical gauges matter as much as the money. coefficient_of_variation and standard_error_percent_change attach sampling precision to each estimate, so a thin industry's print can be weighed before it is quoted - and estimates whose variability exceeds publication standards are suppressed outright rather than shipped as noise. revenue_by_class_of_customer splits demand between governments, businesses and household consumers, and total_expenses covers the not-for-profit-heavy lines where revenue alone misleads. Full definitions sit in the dictionary above.

Three families. First, the seasonally_adjusted_flag, the marker separating the adjusted series from the raw one so both vintages share a single schema instead of doubling the column count. Second, the benchmark keys - category, data type, error type, geography and time-period identifiers - that index the Historical Annual Benchmark Dataset, the machine-readable extract running unbroken from Q4 2003. Third, the annual benchmark extracts themselves, which reach further back than any single quarterly table and cover every selected service industry, not only NAICS 56.

They sit behind an explicit request note rather than padding every delivery because most questions need only one vintage at a time. Name what the analysis requires when requesting the sample and the extract arrives carrying exactly those columns.

Geographically the coverage is strictly United States national totals - no state, metro or county cuts anywhere in the survey. The dataset answers how big and how fast, never where exactly; regional questions need a companion record such as Quarterly Census of Employment and Wages (QCEW) — NAICS 56, which goes the other way on geography.

Temporally the rolling quarterly series reaches to Q4 2003 across the survey's covered industries, with NAICS 56 measured since Q1 2009 - seventeen years of uninterrupted service-sector readings by August 2026. Estimates publish in an advance vintage roughly fifty days after quarter end and a fuller vintage around seventy-five days later, and the annual benchmark against the Service Annual Survey trues the path up once a year.

Granularity stacks in tiers: employer firms aggregated to the NAICS 56 sector total, split into the 561 and 562 subsectors, then resolved to selected industries including 5613 employment services and 5615 travel arrangement and reservation services. Roughly ninety industry rows repeat per quarterly table across all covered sectors, so the whole service economy lands as one flat, joinable frame.

Against the wider Datadory catalog - average quality score 7.81 across all 1,744 datasets - this record scores 7/10, carried by verified field definitions and a GDP-grade mandate, and held back by the hard national-only geographic ceiling.

API, files, or your warehouse. Daily, weekly, or hourly.

Your cadence runs independently of the quarterly release rhythm. Land each vintage the moment it matters to your pipeline - a one-time pull of the full Q4 2003-forward history, or a warehouse table kept current so each new quarter diffs cleanly against the last. What arrives is normalized to the field dictionary above: NAICS codes and kind-of-business labels typed identically whether the row is the sector total or a four-digit industry line, so joins hold without parsing free text.

Deliveries travel with the full field dictionary, the variability tables that say which estimates carry weight, and coverage notes mapped to whatever NAICS cut you named - so the schema you see in the sample is the schema you ship against.

• TAM sizing for support-services markets - staff a market model for staffing, janitorial, security or waste services off the measured $317.8 billion NAICS 561 print instead of a round-up; consultants get a federal citation behind every figure.
• Public-company benchmarking - set a staffing firm's or facility-services provider's reported top line against the industry aggregate so company execution separates from sector tailwind; investors and quants lead fit here.
• Nowcasting and macro-feature pipelines - a series that enters GDP directly is the definition of a load-bearing exogenous feature; data scientists key it on the fixed NAICS identity that joins cleanly to other federal series.
• Demand-composition reads - the class-of-customer split shows whether growth in support services comes from business outsourcing, government contracting or household spend.
• Seasonality forensics - comparing adjusted and not-seasonally-adjusted vintages of the same quarter separates calendar effects from real movement before anyone quotes a number.
• Economic journalism and coursework - cite the standard federal read on the services economy, with methodology documents behind every figure.

Investors & Quants and Market Researchers & Consultants tie at 3/3. The first benchmark reported company top lines against Census quarterly industry totals; the second anchor TAM models on a measured base that reconciles to GDP. Data Scientists & ML Engineers (2/3) get a long, NAICS-keyed quarterly panel built for feature pipelines; Journalists, Academics & Students (2/3) get a citable standard read on the services economy. Sales & Growth Teams and Competitive Intel Product Teams (1/3 each) draw narrower value - sector-momentum talking points and trajectory context respectively - because the file carries no account-level records.

Four things worth settling upfront.

First, geography stops at the national line. No state, metro or county cuts exist in the survey, so any regional question needs a companion dataset beside this one - QCEW's NAICS 56 tabulation is the natural pairing, trading timeliness for geography.

Second, suppression is part of the design. Estimates whose coefficient of variation exceeds publication standards are withheld, so a thin industry can disappear from individual quarters. A gap means too variable to publish, not zero - treat missing rows as information rather than errors.

Third, vintage choice matters. Advance figures land first and get revised; analyses that quote point-in-time values should pin the vintage while analyses that want best-current estimates should take the revised series. Both share one schema under the requested adjusted flag.

Fourth, the identity is NAICS-keyed end to end. Rows join to employment, wage and establishment records on the same codes, which makes this record the revenue half of any NAICS 56 panel assembled from our Diversified Support Services shelf. Name the cut - sector, subsector or industry lines, adjusted or raw, full history or a window - and the sample ships shaped to it.