Retail sales report: what it measures and how to read it

Retail sales measure how much US stores, online sellers, gas stations, car dealers, restaurants and bars took in from customers in a month. The figure comes from the Census Bureau's Advance Monthly Retail Trade Survey, released about two weeks after the month ends, which makes it one of the earliest hard readings on consumer spending, the largest part of the economy.

Markets watch it closely because it arrives before the monthly consumer spending data and long before quarterly GDP. But the headline number has two traps. It is measured in current dollars, so part of any increase is just higher prices. And it is an early estimate from a sample of firms, with a margin of error often as large as the change everyone is reacting to.

This guide explains how the report is built, what it includes and excludes, how to adjust it for inflation, the "control group" subtotal analysts compute, and how to judge whether a monthly change means anything.

What the report measures

The report estimates sales by kind of business for retail and food services firms located in the United States, organized by industry code:

• Motor vehicle and parts dealers (new and used car dealers, auto parts stores).
• Store retailers: furniture, electronics, building materials and garden supply, grocery, health and personal care, gasoline stations, clothing, sporting goods, general merchandise (department stores, warehouse clubs, supercenters) and miscellaneous stores.
• Nonstore retailers, mostly online sellers and mail order.
• Food services and drinking places: restaurants and bars, the only service industry in the total.

It classifies businesses, not products. A grocery chain's sales of TVs count as grocery store sales; an online retailer's sales of groceries count as nonstore.

How the advance estimate is built

The Census Bureau runs two linked surveys. The full Monthly Retail Trade Survey (MRTS) covers a large sample. The Advance Monthly Retail Trade Survey (MARTS) is a subsample: each month the Bureau mails a probability sample of about 4,800 employer firms, whose sales are weighted and benchmarked to represent the complete universe of more than three million retail and food services firms.

For each detailed industry the Bureau computes the ratio of this month's to last month's sales among firms that reported both months, then applies that ratio to last month's estimate from the larger survey. Estimates are seasonally adjusted with the X-13ARIMA-SEATS program, and adjusted for holiday and trading-day differences: a month with five Saturdays is not the same as one with four.

When it comes out

The advance report is published at 8:30 a.m. Eastern, about two weeks after the month ends. The September 2026 report came out on October 15, 2026; October's is scheduled for November 17, 2026. Each report also revises the previous two months, using the larger survey, and the full history is revised once a year: the latest annual revision was released on September 28, 2026.

Reading a report: September 2026

A dollar-sized monthly sales figure looks odd at first: $745 billion is one month of sales, not an annual rate. Multiplied by 12, it is about $8.9 trillion a year, against total personal consumption expenditures of about $22.5 trillion (annual rate) in the same month. Retail sales therefore cover roughly 40% of consumer spending. Housing, health care, financial services, education, transport fares and most other services are outside the survey. That is why a strong retail month can coexist with weak overall consumption, or the reverse.

The subtotals: ex autos, ex gas and the control group

Two industries swing the total for reasons unrelated to underlying demand:

• Auto dealers sell large-ticket items whose sales jump with model-year incentives, supply disruptions or storms that destroy cars.
• Gasoline stations sell a product whose price moves daily. A rise in pump prices raises their sales even if drivers buy fewer gallons.

The report publishes totals excluding each of them. Analysts also compute a narrower subtotal, usually called the control group: total sales minus autos, gasoline stations, building materials and food services. It strips out the most volatile pieces and the restaurant line, leaving goods spending that tracks the goods part of consumer spending closely.

Most of the gap came from gasoline. Sales at gas stations were $68.0 billion, up 28.1% from $53.1 billion a year earlier, while the total excluding gasoline stations was up 4.5%. The Census Bureau does not publish confidence intervals for analysts' subtotals, so a 0.14% move in the control group should be read as roughly flat.

Nominal, not real

The release says it plainly: the data are adjusted for seasonal variation and for holiday and trading-day differences, "but not for price changes". A 6.3% annual rise in sales with prices up 3.7% is not 6.3% more stuff.

The chart shows how much inflation has flattered the headline. From January 2015 to September 2026 nominal sales rose 74%, from $428.2 billion to $745.0 billion a month. Deflated by CPI, the rise was about 21%. Since February 2020 nominal sales are up 44%, while consumer prices rose 30%; the real gain is about 11%. In inflation-adjusted terms, sales peaked in April 2022 at about $770.9 billion (in September 2026 dollars) and in September 2026 were still about 3.4% below that level.

Deflating by the all-items CPI is a rough method. CPI includes rent, insurance and other services that retailers don't sell, and goods prices often rise more slowly or even fall while service prices climb. A more careful deflator uses goods price indexes matched to each store type, which is roughly what BEA does when it turns retail sales into real consumer spending. Still, the simple adjustment is enough to stop a reader from mistaking inflation for demand.

A real example from the data: 2020 and 2021

The pandemic produced the largest swings in the series:

• February to April 2020: sales fell from $515.8 billion to $402.2 billion, a drop of 22.0% in two months.
• May 2020: sales jumped 19.4% in a single month; by June they were 0.6% above February's level.
• April 2021: sales were 51.5% higher than in April 2020, the biggest annual gain in a series that starts in 1992. Most of that was a base effect: April 2020 was the low point. Compared with April 2019, the rise was 21.9%.

The lesson for year-over-year growth rates is general: a change from an unusual month tells you as much about the base as about the present.

How markets read retail sales

Retail sales feed directly into estimates of consumer spending and therefore GDP, so a surprise can change forecasts for the quarter within minutes. As with other releases, markets react to the difference from consensus rather than to the level; see How markets react to economic data: surprises, regimes, revisions.

What traders look at beyond the headline:

• The control group and ex-autos-and-gas totals, to strip out price and auto swings.
• Revisions to the prior two months, which can outweigh the new month.
• The confidence interval. A 0.3% surprise against a ±0.5% interval is within noise.
• Gasoline's share of the change, which says more about oil prices than about consumers; see Oil prices and the economy: WTI, Brent, gasoline and inflation.
• Restaurant sales, the only services line, sometimes read as a gauge of discretionary spending.

Strong sales tend to push Treasury yields up, since they point to faster growth and less room for rate cuts. For stocks, the effect depends on whether investors fear a slowdown or inflation. Retailers and consumer discretionary stocks can also move on the details, such as nonstore against department store sales.

Pitfalls and limitations

• It's nominal. In an inflationary year a 5% rise can mean flat volumes. Compare with CPI before drawing conclusions.
• Gasoline and car prices dominate months. Check the ex-autos and ex-gas totals, or build the control group, before reading the headline.
• Sampling error. The advance estimate comes from about 4,800 firms. Changes inside the 90% confidence interval are not distinguishable from zero.
• Revisions. The advance number is revised twice in the next two reports and again each year; the August 2026 change moved from +1.1% to +1.3% in a month.
• Goods-heavy. Services make up most consumer spending and are mostly outside the survey. Retail sales can fall while total spending rises.
• Seasonal adjustment is an estimate. Holiday timing, Easter and weather can leave distortions, especially in December and around spring holidays; see Seasonal adjustment and data revisions: why the numbers change.
• Business, not product. Shifts between channels, such as from department stores to online sellers, change the composition without changing total demand.

Retail sales compared with other spending measures

Retail sales come first and cover a large slice of goods spending. The broader monthly consumption figures from BEA follow about two weeks later and fold retail sales into a complete, inflation-adjusted picture. Surveys of how consumers feel are a different kind of data, covered in Consumer sentiment and confidence: how the surveys work. The quarterly picture is in GDP explained: how gross domestic product is measured and read.

How Finabase does it

Finabase does not publish retail sales data or an economic calendar. The chart and the figures on this page are ours, built from Census Bureau and BLS data distributed through the St. Louis Fed's FRED database and from the Census Bureau's September 2026 release.

What the app shows is the market side. The Consumer discretionary sector page lists the sector's S&P 500 stocks, including retailers, with their sector index, and the Sector page compares all 11 sectors' returns. Retail ETFs such as the SPDR S&P Retail ETF have their own pages, for example XRT, with quotes and charts. The Discover page's Economy topic collects web articles about the economy, including retail sales releases.

Frequently asked questions

Are retail sales adjusted for inflation?

No. The Census Bureau adjusts them for seasonal variation and holiday and trading-day differences, but not for price changes. To estimate real growth, divide one plus the nominal growth rate by one plus the inflation rate: 6.3% sales growth with 3.7% CPI inflation is about 2.5% real growth.

When are retail sales released?

The advance report comes out at 8:30 a.m. Eastern about two weeks after the month ends. September 2026 data were released on October 15, 2026, and October data are scheduled for November 17, 2026.

What is the retail sales control group?

It is total retail and food services sales minus auto dealers, gasoline stations, building materials stores and food services. Analysts compute it because it removes the most volatile lines and tracks the goods part of consumer spending.

Do retail sales include online shopping?

Yes. Online and mail-order sellers are counted as nonstore retailers: $124.3 billion in September 2026, seasonally adjusted, about one-sixth of the total.

Do retail sales include services?

Only restaurants and bars. Rent, health care, utilities, travel and most other services are not part of the survey, which is why retail sales cover only about 40% of consumer spending.

Summary

The Census Bureau's advance retail sales report estimates a month's sales at stores, online sellers, car dealers, gas stations and restaurants from a sample of about 4,800 firms, roughly two weeks after the month ends. It is seasonally adjusted but not adjusted for inflation, so growth must be compared with price changes: September 2026's 6.3% annual gain was about 2.5% after inflation. Analysts strip out autos, gasoline, building materials and restaurants to get the steadier control group, check the confidence interval and revisions before believing a monthly move, and remember that retail sales cover only about 40% of consumer spending.