Guide

Quarterly services survey explained

Harbor Hospitality runs 85 full-service hotels and a corporate travel desk. In Q1 2024 their revenue operations team built a GDP nowcast from retail sales control group plus ISM Services PMI diffusion indexes. The model predicted +2.8% annualized real services PCE growth. BEA’s first print came in at +1.9%. Post-mortem: retail sales captured goods and restaurant tabs but missed finance, healthcare, and transport revenue — roughly 70% of Harbor’s corporate segment exposure. After adding Census Quarterly Services Survey (QSS) revenue by NAICS sector, the same nowcast’s Q1 tracking error fell from 90 basis points to 24 bps on a three-quarter rolling average.

The Quarterly Services Survey is the Census Bureau’s revenue census for selected service industries. Unlike diffusion surveys that ask “better or worse than last month,” QSS collects dollar revenue from establishments — seasonally adjusted levels and quarter-over-quarter percent changes by sector. It is slower than monthly retail sales but broader: professional services, information, finance, healthcare, transportation, and administrative support appear here, not in advance retail trade. This guide covers what QSS measures, covered NAICS sectors, methodology and revisions, how it bridges to BEA PCE services and GDP, the Harbor Hospitality refactor, a technique decision table, pitfalls, and a production checklist.

What the Quarterly Services Survey measures

QSS estimates total operating revenue for service-providing firms in the United States. Census mails or electronically collects reports from a probability sample of establishments; results are weighted to represent the full population of covered industries. Published figures are nominal — not inflation-adjusted — and seasonally adjusted at the sector level.

Core published series

Total revenue for all QSS sectors — aggregate headline.
Revenue by NAICS sector — e.g., information, finance and insurance, health care, transportation and warehousing, professional/scientific/technical services.
Quarter-over-quarter percent change — primary growth read; also year-over-year for trend context.
Selected subsector detail — varies by release vintage; Census expands or consolidates lines as sample redesigns land.

QSS is quarterly, typically released about 75 days after quarter-end (roughly mid-February, May, August, and November). That lag trades timeliness for breadth — you get hard revenue dollars across services retail sales cannot see.

Dimension	QSS	Retail sales (goods)
Frequency	Quarterly (~75-day lag)	Monthly (~2-week lag)
Unit	Nominal revenue dollars	Nominal sales dollars
Coverage	Selected service industries	Retail establishments (mostly goods + restaurants)
Method	Establishment revenue census	Advance monthly sample

Covered industries and what is excluded

QSS does not cover the entire economy. Census selects NAICS sectors where establishment revenue is measurable and economically important for GDP compilation. Typical inclusions:

Information — publishing, motion picture, broadcasting, telecommunications, data processing and hosting.
Finance and insurance — banking fees, securities commissions, insurance premiums (revenue to carriers, not claims paid).
Real estate and rental and leasing — lessor revenue; owner-occupied housing imputation still lives in BEA, not QSS.
Professional, scientific, and technical services — legal, accounting, consulting, architecture, engineering.
Administrative and support services — employment services, travel arrangement, security, facilities support.
Health care and social assistance — hospital and ambulatory revenue; government-run facilities may be partial.
Arts, entertainment, and recreation — venues, gambling, fitness, museums.
Accommodation and food services — hotels and restaurants (restaurants also appear in monthly retail sales).
Transportation and warehousing — air, rail, truck, couriers, warehousing revenue.
Other services — repair, personal care, religious and civic organizations where sampled.

Common exclusions

Goods-producing sectors — manufacturing, mining, construction revenue is elsewhere (manufacturing shipments, construction spending).
Government non-enterprise activity — public education and much of Medicare/Medicaid flows are compiled through administrative records, not QSS establishment forms.
Small-firm cutoffs — sample thresholds omit some micro-establishments; coverage ratios differ by sector.

When a sector line moves sharply, check whether Census revised NAICS definitions or sample frames — structural breaks are rarer than in retail sales but do happen on five-year economic census cycles.

Methodology, seasonal adjustment, and revisions

QSS uses a stratified random sample of employer establishments. Large firms may report for all locations; smaller firms represent a slice. Nonresponse is imputed from matched firms with similar size and industry. Seasonal adjustment removes predictable quarter effects (summer travel, year-end financial fees, tax-season accounting).

Revision schedule

Advance quarterly estimate — first publication ~75 days after quarter-end.
Revised estimate — published with the next quarter’s advance; incorporates late responses and improved seasonal factors.
Annual benchmark — tied to the Service Annual Survey (SAS) and Economic Census; can re-level several years of history.

Unlike monthly retail sales, you get only four observation points per year — so a single revised quarter can flip a whole-year narrative. Track both advance and revised vintages if you feed QSS into models.

Nominal vs real

QSS revenue rises when prices rise even if transaction volumes are flat. To estimate real services growth, deflators come from BEA (PCE services price indexes) or BLS PPI for specific subsectors. A +4% nominal QSS print with +3% services inflation implies roughly +1% real volume — back-of-envelope; sector deflators differ.

QSS vs ISM Services PMI and other high-frequency proxies

Markets trade ISM Services PMI on the first business day of each month. PMI is a diffusion index: share of firms reporting expansion minus contraction. It leads sentiment and employment direction but does not measure revenue dollars.

Signal	Strength	Weakness
ISM Services PMI	Monthly, timeliness, market-moving	No dollar levels; qualitative; sector mix differs from GDP weights
QSS revenue	Hard nominal revenue by NAICS; BEA input	Quarterly lag; partial sector coverage
Retail sales (food services line)	Monthly restaurants and bars	Tiny slice of total services PCE
Card-spending aggregates (private)	Weekly granularity	Sample bias, classification opacity, not official benchmark

A sound macro desk pairs monthly diffusion (ISM, S&P Global PMI) with quarterly revenue (QSS) when reconciling services PCE ahead of BEA’s first GDP vintage. Diffusion can roll over three months before QSS revenue growth slows — or diverge when price inflation inflates nominal revenue while activity softens.

Bridge to PCE services and GDP

Personal consumption expenditures on services is the largest block of U.S. GDP — healthcare, housing services, financial services, recreation, and transport. BEA combines many inputs: QSS where coverage aligns, administrative data (CMS for healthcare, IRS for some fees), and indicator extrapolations between QSS vintages.

Information and professional services — QSS revenue trends inform BEA’s quarterly extrapolation before annual benchmarks.
Accommodation and food — cross-check monthly retail sales food-services line against QSS quarterly totals.
Finance and insurance — QSS captures fee revenue; BEA separates imputed services (owner-occupied rent, free banking) that QSS never sees.
Healthcare — QSS hospital and ambulatory revenue complements CMS claims data; government programs add off-survey flows.

GDP nowcasters often regress quarterly changes in BEA real services PCE on lags of QSS nominal growth, ISM business activity, and retail control group — then apply BEA’s latest services deflator assumptions. QSS revisions can move the entire quarter’s GDP tracking error when BEA trues up to the Service Annual Survey.

Worked example: Harbor Hospitality nowcast refactor

Harbor Hospitality’s corporate segment sells negotiated room blocks and event packages to finance, consulting, and healthcare clients. Their old nowcast overweighted goods consumption proxies. The refactor added QSS sector revenue growth rates with explicit NAICS mapping:

Map client exposure to QSS lines — finance and professional services revenue growth weighted 40%; accommodation QSS line 25%; transportation 15%; health care 10%; arts/entertainment 10%.
Pull advance QSS quarter-over-quarter changes on release morning; store revised vintage when next quarter publishes.
Deflate with BEA services PCE deflator (lag one month from personal income report) for real growth estimate.
Blend with monthly ISM Services business activity for intra-quarter bridge — ISM gets 30% weight only inside the current quarter before QSS prints.
Cross-check retail food services for restaurant-heavy markets; flag divergence >150 bps between QSS accommodation and retail food line.
Log tracking error vs BEA first and third GDP vintages — Harbor’s three-quarter MAE fell from 90 bps to 24 bps on real services PCE after QSS integration.

The team still trades around monthly PMI and payrolls for tactical hedges, but quarterly capital planning for new properties now keys off QSS professional-services and accommodation revenue trends rather than goods-only retail sales.

Technique decision table

Question you have	Best indicator	Why not the alternative alone
Monthly services momentum for rates day	ISM Services PMI + retail food services	QSS is too stale for FOMC-week positioning
Quarterly services PCE nowcast	QSS sector revenue + BEA deflator	PMI diffusion has no dollar scale
Goods vs services consumption split	Retail control group + QSS total	Headline retail sales misses most services
Healthcare services volume trend	QSS health care revenue + CMS utilization	Retail sales has no healthcare line
Travel and hotel cycle	QSS accommodation + retail food services	ISM alone cannot quantify room-night revenue
Financial services fee income cycle	QSS finance and insurance revenue	Stock market levels are not fee revenue
Real volume vs price in services	QSS nominal + PCE services deflator	Nominal QSS alone confuses inflation with growth

Common pitfalls

Treating QSS as total GDP services output — coverage is partial; BEA adds imputations and government-compiled health data.
Ignoring revisions — advance QSS can flip sign on revision; do not lock models to first print only.
Comparing to real PCE without deflating — nominal QSS acceleration during inflation spikes is not volume strength.
Double-counting restaurants — food services appear in both retail sales and QSS; pick one bridge or net explicitly.
PMI–QSS divergence without diagnosis — diffusion can fall while nominal revenue rises on price; decompose before calling recession.
Quarterly noise in small sectors — arts/entertainment and repair lines are volatile; use four-quarter averages.
Missing benchmark years — Economic Census years reset levels; backtests must handle level shifts.
Using QSS for high-frequency trading — 75-day lag makes it structural, not tactical; pair with monthly data.

Production checklist

Download advance and revised QSS vintages from Census; version-control both.
Build NAICS-to-PCE mapping table aligned with BEA industry correspondence.
Store release calendar (mid-Feb, May, Aug, Nov) with BEA GDP print offsets.
Deflate nominal QSS sectors with matching PPI or PCE deflator subcomponents.
Reconcile QSS accommodation with retail food services monthly.
Log ISM–QSS divergence flags when PMI rolls for two months but QSS growth holds.
Re-estimate nowcast coefficients after annual Service Annual Survey benchmark.
Document excluded sectors so consumers know what QSS does not cover.
Archive revised history; do not overwrite advance-only series silently.
Cross-read GDP nowcast methodology when blending QSS with goods indicators.

Key takeaways

QSS is Census’s quarterly revenue census for selected service industries — nominal dollars, not diffusion sentiment.
It fills the gap retail sales leaves: finance, healthcare, transport, and professional services dominate U.S. consumption.
BEA uses QSS as a key input to services PCE; nowcasts improve when sector revenue replaces PMI-only extrapolation.
Advance vs revised vintages and annual benchmarks matter — treat first print as provisional.
Pair quarterly QSS with monthly ISM and retail food services for timeliness without losing dollar grounding.