Inside the BTOS: How a Biweekly Survey of 1.2 Million Businesses Provides Real-Time Economic Data

Every two weeks, a quiet pulse check goes out to 200,000 American businesses. It arrives via email, asking a standardized set of questions about revenue, employment, and prices over the previous 14 days. This is the U.S. Census Bureau's Business Trends and Outlook Survey (BTOS), a continuous instrument designed to track the real-time health of the economy.[1]

The scale of the endeavor represents a structural shift in how the federal government measures economic reality. The total BTOS sample consists of approximately 1.2 million businesses, split into six rotating panels. By replacing slower, heavier quarterly surveys with a high-frequency biweekly cadence, the Census Bureau aims to bridge the gap between anecdotal sentiment and hard, lagging macroeconomic indicators.[1]

The mechanism is built for speed rather than exhaustive accounting. Businesses in each of the six panels are asked to report once every 12 weeks for a year, meaning a single business will complete the survey four to five times before rotating out of the sample. The Census Bureau estimates that it takes the average respondent approximately nine minutes to complete the questionnaire, which relies on qualitative estimates rather than audited financial records.[1]

The survey tracks core economic phenomena by measuring extensive margins—the share of businesses engaging in an activity, rather than the dollar-value intensity of that activity. Topics include current performance, changes in revenue, hours worked, demand, and prices. This allows the BTOS to provide local, state, and federal officials with near real-time data for policy and decision-making, particularly in the wake of natural disasters or sudden economic crises.[1]

Because the underlying sample is so massive, the data can be parsed with remarkable geographic and sectoral granularity. Results are published every two weeks and are available by 2022 North American Industry Classification System (NAICS) sector, by state, and for the 25 most populous Metropolitan Statistical Areas. This localized view captures the immediate impact of regional shocks and assists in monitoring recovery efforts.[1]

Beyond the core economic questions, the BTOS serves as a rapid-deployment vehicle for emerging trends. The Census Bureau periodically adds supplemental content to the survey to measure phenomena that traditional instruments miss. Recent supplements have tracked the adoption of Artificial Intelligence (AI) and the prevalence of Work From Home (WFH) arrangements across the U.S. economy.[1][2]

A recent analysis of BTOS data by researchers at the University of Chicago pooled responses gathered during a 12-week fielding cycle to examine WFH metrics. The resulting dataset covered over 150,000 businesses, providing a highly detailed look at remote work practices from the employer's perspective.[2]

The evidence reveals massive sectoral heterogeneity. While employees work from home about one day per week on average across all firms, the Information sector averages 2.8 remote days per week—twenty times higher than the 0.1 days per week observed in the Accommodation and Food Services sector. The data confirms that fully remote work remains concentrated in specific knowledge-work industries.[2]

Similarly, the BTOS AI supplement provided a reality check against market hype. By surveying the 1.2 million-business sample, the Census Bureau found that AI adoption is highly uneven, with larger firms showing significantly higher usage rates than small businesses. The biweekly cadence allows policymakers to watch this technology diffuse through the economy in real time, rather than waiting years for retrospective census data.[1]

However, the speed and scale of the BTOS come with explicit methodological trade-offs. The most significant limitation is the response rate. According to the University of Chicago analysis, the survey averages a 13% response rate over a 12-week cycle. While a 13% yield on a 1.2 million-business sample still produces a massive absolute number of responses—roughly 150,000 per cycle—the 87% non-response rate introduces potential selection bias.[2][3]

The evidence is thin on how non-respondents differ from those who complete the survey. Businesses under extreme financial stress, those lacking administrative staff, or those that have recently closed may simply ignore the email invitations. If struggling firms are less likely to respond, the resulting data could skew slightly optimistic, painting a rosier picture of the economy than reality dictates.[3]

Despite the low response rate, the aggregate effort required to maintain this real-time tracking is substantial. Based on the active sample size of 1.2 million businesses, the annualized frequency of panel rotation (4.33 times per year), the 13% observed response rate, and the agency's estimated 9-minute completion time, Factlen calculates that the BTOS requires approximately 101,400 hours of actual respondent time annually.[1][2][3]

While 101,400 hours is a significant aggregate figure, it is spread thinly across the U.S. economy. For the individual business owner, the burden is roughly 36 to 45 minutes per year. This represents a remarkably low aggregate cost for high-frequency national data, especially compared to the heavy administrative burden of traditional mandatory economic censuses.[1][3]

Research published by the American Economic Association highlights how the BTOS successfully captures nuanced sectoral differences despite its qualitative nature. For example, the data shows that businesses in the Accommodation and Food Services sector frequently anticipate future output price increases, likely as a necessary mechanism to accommodate rising current input prices.

This qualitative approach—asking whether prices are "higher," "lower," or "the same"—allows the survey to move faster than instruments requiring exact dollar figures. It trades the precision of audited financials for the immediacy of directional trends, giving the Federal Reserve and other policymakers an early warning system for inflationary pressures or demand shocks.[3]

As the global economy moves faster and becomes more susceptible to rapid shocks—from pandemics to supply chain snarls—the need for high-frequency data will only grow. The BTOS represents a permanent evolution in economic methodology. By continuously polling 1.2 million businesses, the U.S. government has built a real-time dashboard for the American economy, accepting the inherent uncertainties of rapid surveying in exchange for unprecedented speed.[1][3]