Retail Sales Report: Reading the Data

Table of Contents

    The retail sales report can show spending rising while the underlying consumer signal is much less impressive. Higher gasoline prices can lift gas-station receipts, a jump at auto dealers can dominate the total, and a revision can change the starting point for the latest month’s growth. Reading only the headline misses these distinctions.

    This guide explains the U.S. Census Bureau’s release, the differences among headline and commonly watched core measures, and how to connect the figures to consumer spending without treating them as a direct forecast of real GDP.

    What Is the Retail Sales Report?

    The Census Bureau’s Advance Monthly Sales for Retail and Food Services report provides an early national estimate of sales at retail and food service establishments in the United States. The advance survey, known as MARTS, is based on a sample of employer firms. It is generally released about ten business days after the reference month and revised a month later using the larger Monthly Retail Trade Survey, or MRTS. Actual release dates should be checked on the Census calendar. [1]

    Three Questions Before Interpreting the Number

    Which businesses are included? Is the change adjusted for calendar patterns or prices? And which release vintage supplies the current and previous month’s values? A growth rate is useful only after those choices are clear.

    The report measures sales dollars, rather than physical quantities, household purchasing power or retailer profits. Its categories describe the seller’s kind of business. Gasoline-station sales, for example, are receipts at those establishments rather than a count of gallons sold everywhere. Census specifically cautions that sales by type of store are not total sales of a particular product across the economy. [2]

    Coverage also matters when comparing historical downloads. In the April 2025 annual revision, Census removed nonemployer firms, meaning businesses without paid employees, from the estimates and historical time series to align with the Annual Integrated Economic Survey. They had previously been included through a benchmarking adjustment. Compare observations from a consistent revised series rather than splicing an older release’s level into a current download. [1]

    Headline vs Core and Control-Group Sales

    The headline generally refers to total retail and food services sales. “Retail trade” alone is narrower because it excludes food services. Other measures remove particular business categories from the headline total. The exclusions must be stated: “ex-autos,” “ex-gas” and “control group” answer different questions.

    Common measures derived from the retail and food services total
    Measure What is removed? What it helps isolate
    Headline None of the categories below The broad retail and food services sales total
    Ex-autos Motor vehicle and parts dealers Sales outside that dealer category; gasoline stations remain
    Ex-gas Gasoline stations Sales outside gas stations; motor vehicle and parts dealers remain
    Ex-autos and gas Both categories above A total that still includes building-material dealers and food services
    Common control-group measure Motor vehicle and parts dealers; gasoline stations; building material and garden equipment and supplies dealers; food services and drinking places A narrower goods-oriented spending signal often used in consumption analysis

    Census publishes several exclusion totals, but it does not designate a universal official “core retail sales” measure. A Richmond Fed explanation uses core for the four-exclusion measure in the final row. A data calendar or commentary may use the word differently, so inspect the definition rather than assuming that “core” means the same thing everywhere. [3] [9]

    Removing a category does not make it economically unimportant. BEA has other sources for estimating consumer purchases of vehicles and motor fuels. Some building-material purchases are associated with residential investment, and restaurant receipts still inform PCE services even though the common market control-group measure excludes them. The exclusions help with a particular statistical bridge; they do not identify spending that can be ignored. [3] [7]

    A Worked Retail Sales Example

    Consider a hypothetical simplified retail economy with five nonoverlapping groups. All figures below are invented sales amounts in billions of U.S. dollars, already seasonally adjusted. Month 0 is the comparison month and Month 1 is the new month. Prices have not been removed.

    Hypothetical category sales, USD billions
    Business category Month 0 Month 1
    Motor vehicle and parts dealers 20.0 22.0
    Gasoline stations 10.0 11.0
    Building material and garden dealers 10.0 10.0
    Food services and drinking places 10.0 10.5
    Remaining retail categories 50.0 50.5
    Headline total 100.0 104.0

    For any chosen measure, first add the retained category sales in each month. Then calculate the percentage change in that aggregate:

    Monthly Sales Growth
    Growth (%) = [(SalesMonth 1 / SalesMonth 0) − 1] × 100
    Use the same coverage, seasonal-adjustment basis and release vintage in both months.
    The same hypothetical sales produce different growth rates
    Measure Retained sales: Month 0 → Month 1 Calculation
    Headline 100 → 104 (104 / 100 − 1) × 100 = 4.00%
    Ex-autos 80 → 82 (82 / 80 − 1) × 100 = 2.50%
    Ex-gas 90 → 93 (93 / 90 − 1) × 100 = 3.33%
    Ex-autos and gas 70 → 71 (71 / 70 − 1) × 100 = 1.43%
    Common control group 50 → 50.5 (50.5 / 50 − 1) × 100 = 1.00%

    The headline’s 4.00% increase is much stronger than the remaining retail categories’ 1.00% increase. Excluding autos and gas does not reach the control-group result because it still includes the restaurant increase and unchanged building-material sales. None of these five rates measures inflation-adjusted growth.

    Do not subtract an unweighted category growth rate from headline growth. A 10% increase in a small category has a different effect from the same increase in a large category. Calculate with dollar levels or appropriate weights.

    A real release can also show different signals. In the May 16, 2023 release for April 2023, headline sales rose 0.4% from the previous month. The Richmond Fed subsequently described the common four-exclusion core measure as rising 0.7%. Those are contemporaneous historical readings, not today’s revised estimates. [8] [3]

    Nominal vs Real and Seasonally Adjusted Sales

    The headline monthly comparison is adjusted for recurring seasonal variation and holiday and trading-day differences, but not for price changes. Seasonal adjustment helps compare months with different normal shopping patterns. It does not convert receipts into quantities or remove inflation. [1] [9]

    For example, a decline from December to January in unadjusted receipts may reflect the usual end of holiday shopping. The seasonally adjusted change asks whether sales were stronger or weaker than the recurring calendar pattern would suggest. For the broader statistical issue, see seasonality in time-series regression.

    Price adjustment asks a separate question: how much did the amount purchased change after allowing for prices? In a separate hypothetical fixed-composition basket, suppose nominal spending rises from 100 to 104 while its matching price index rises from 100 to 103.

    Simple Real Growth With a Matching Price Index
    Real growth (%) = [(1 + gnominal) / (1 + gprices) − 1] × 100
    The growth inputs are decimals over the same period and for the same basket.
    Higher Receipts, Much Smaller Real Growth

    Nominal growth is (104 / 100 − 1) × 100 = 4%. Matching price growth is (103 / 100 − 1) × 100 = 3%.

    Real growth is (1.04 / 1.03 − 1) × 100 = 0.9709%, or approximately 0.97%.

    Simply subtracting 3% from 4% gives an approximation of about 1% real growth. The exact ratio is slightly different. More importantly, subtracting the broad Consumer Price Index from total retail-sales growth does not reproduce official real PCE: the baskets, category weights and aggregation methods differ. BEA generally uses appropriate detailed price indexes and combines components with chain-type measures. [6] [7]

    The same conceptual distinction appears in real versus nominal GDP. A larger dollar total can reflect more purchases, higher prices, a different mix of purchases, or a combination of these effects.

    How Revisions and Sampling Uncertainty Change the Story

    The advance estimate is an early sample-based estimate. Later information from the larger MRTS and annual benchmarking can change previously published values. Read the revision alongside the new month, and retain the release date when comparing a forecast with what was known at the time. Today’s revised history is not the information investors originally received. [1]

    A Revision Can Change the Growth Rate’s Starting Point

    In a separate hypothetical revision example, originally reported sales rise from 100 in the previous month to 101 in the latest month. That implies (101 / 100 − 1) × 100 = 1.00% growth. If the previous month is revised to 102 while the latest month stays at 101, the comparison becomes (101 / 102 − 1) × 100 = −0.98%.

    The sign changed because the denominator changed. In actual releases both months can change, so compare the revised levels and rates rather than attributing every change in the story to new activity.

    The archived April 2023 report provides a real revision example: it revised the February-to-March change from −0.6% to −0.7%. That is a downward revision of −0.7 − (−0.6) = −0.1 percentage point, not a 0.1% decline in the dollar level. [8]

    A Positive Estimate May Still Be Statistically Uncertain

    In that same release, April’s headline increase was 0.4% with a ±0.5 percentage-point sampling margin at the 90% confidence level. The interval was therefore 0.4 − 0.5 = −0.1% to 0.4 + 0.5 = 0.9%. It included zero, as the release explicitly flagged. [8]

    Uncertain Does Not Mean Unchanged

    An interval containing zero means the survey does not provide sufficient evidence of a nonzero change at that confidence level. It does not prove that sales were flat. The sampling margin also does not measure every possible error: nonresponse, coverage and processing can introduce nonsampling error.

    These historical figures belong to the May 2023 release vintage, which predates the later nonemployer coverage revision. Their purpose here is to show how to read an estimate and its qualifications, rather than to supply a current sales-history comparison.

    Retail Sales vs PCE and GDP

    Personal consumption expenditures, or PCE, is broader than retail sales. It measures goods and services consumed by U.S. resident households and nonprofit institutions serving households, including extensive spending on housing and health care. Census retail sales provide important source information, but they do not capture all those services or match PCE’s concepts exactly. [6]

    BEA’s retail-control method uses retail sales to help estimate PCE for most goods. It extrapolates detailed PCE categories using indicator series, then reconciles the covered aggregate with the retail-control growth rate. The retail and PCE categories need not have the same dollar levels or identical classifications. BEA uses other methods or sources for various remaining goods and services. [4] [5] [7]

    For a simplified hypothetical illustration, a selected retail aggregate rising from 200 to 202 grows by (202 / 200 − 1) × 100 = 1%. Applying that growth to a different covered PCE aggregate starting at 150 gives 150 × 1.01 = 151.5. Equal growth rates do not imply equal levels, and neither number describes all PCE. BEA’s actual method also reconciles the detailed categories.

    The common market control group is therefore a useful clue to part of consumption, rather than a complete PCE estimate. In particular, dropping restaurant sales from that measure does not make restaurant spending disappear from the national accounts: Census food-services receipts inform the relevant PCE services estimates. [7]

    Moving from the release to quarterly gross domestic product requires additional steps: estimate the other consumption categories, account for prices, combine the months into the relevant quarterly measures, and consider the other components of GDP. A monthly nominal retail-sales change is not a quarterly annualized real-PCE growth rate or a percentage-point contribution to GDP growth. Multiplying it by three, four or twelve does not perform that bridge.

    How to Read a Retail Sales Release

    1. Identify the month, release date and basis. Distinguish month-over-month from year-over-year growth, seasonally adjusted from unadjusted data, and advance from revised observations.
    2. Read the headline and the prior-month revision together. Check whether the apparent momentum comes from the new month, a changed starting point or both.
    3. Compare clearly defined measures. Look at dealer and gas-station effects, then inspect the common control-group aggregate and the remaining category detail.
    4. Ask what prices could explain. Larger receipts do not establish larger quantities. Use appropriately matched price and PCE information before drawing real-spending conclusions.
    5. Read uncertainty and coverage notes. A small monthly movement, changed seasonal factors or an annual revision can matter to the interpretation.
    6. Check the wider consumer picture. Compare with real PCE, income and labor-market evidence. The broader business-cycle indicators framework helps organize signals that need not agree in one month.

    For investment analysis, separate the economic reading from the market reaction. A stronger release can change views about sales, inflation and interest rates at the same time. The response also depends on what was already expected. The report alone does not prescribe a direction for stocks or bonds, nor does national sales growth establish any particular retailer’s revenue or profit growth.

    Common Mistakes When Reading Retail Sales

    • Treating “core” as a complete definition. State the excluded business categories, especially whether building materials and food services remain.
    • Calling seasonally adjusted sales real growth. Calendar adjustment and price adjustment solve different problems.
    • Subtracting category growth rates without weights. Exclusion aggregates must be calculated from consistent levels or appropriate weighted contributions.
    • Ignoring revisions and mixing vintages. A revised prior month changes the base; a historical release and today’s time series can legitimately differ.
    • Reading a positive estimate as a precise finding. Check sampling uncertainty and remember the additional nonsampling risks.
    • Turning control-group growth into all-PCE or GDP growth. Coverage, prices and monthly-to-quarterly timing still need to be addressed.

    Limitations of the Retail Sales Signal

    The report is timely, but its scope is selective and its earliest values are revisable. It cannot reveal all consumer services, the distribution of spending across households, how purchases were financed, or whether a sales increase improved retailers’ margins. Store-type totals also do not identify every product purchased across all outlets.

    A single month’s movement can reflect timing, prices, composition and unusual events as well as changing demand. Seasonal adjustment does not remove every irregular influence, and excluding several categories does not eliminate uncertainty in the remaining total. A sustained pattern with corroborating income, employment and real-spending evidence is more informative than one isolated figure.

    Similarly, one weak release does not establish a recession or a numerical probability of recession. Treat retail sales as one piece of evidence whose meaning depends on coverage, persistence and the rest of the economy.

    Frequently Asked Questions

    The common market measure removes motor vehicle and parts dealers, gasoline stations, building material and garden equipment and supplies dealers, and food services and drinking places. It is narrower than sales excluding only autos and gas. Always check the publisher’s definition because “core” is used inconsistently.

    The Census headline is adjusted for seasonal variation and holiday and trading-day differences, but not for price changes. It is a nominal sales measure. Estimating real consumption requires appropriate price information; subtracting broad CPI inflation from total retail-sales growth does not reproduce official real PCE.

    The advance survey provides an early estimate. A larger monthly survey supplies later information, and benchmarking and other revisions can update history. Since growth compares two periods, changes to either period can alter the reported rate. Record the release vintage when comparing estimates over time.

    No. It means the survey does not provide sufficient statistical evidence that the change differs from zero at the stated confidence level. Positive and negative changes can remain consistent with the sampling uncertainty. The interval also does not quantify every source of nonsampling error.

    It can inform the estimate, especially for selected goods consumption, but it is not enough on its own. Quarterly GDP also requires other goods and services, prices, quarterly timing and nonconsumption components. A monthly nominal control-group growth rate cannot be read directly as quarterly annualized real-PCE growth or a GDP contribution.

    Sources

    1. U.S. Census Bureau: Advance Monthly Retail Trade Survey Methodology. Survey timing, employer sample, April 2025 nonemployer removal, seasonal adjustment, benchmarking and uncertainty.
    2. U.S. Census Bureau: Monthly Retail Trade Definitions. Current-dollar values and the distinction between store-type sales and product sales.
    3. John O’Trakoun, Richmond Fed: What Are “Core Retail Sales?”, May 30, 2023. Common four-category exclusions and the contemporaneous April 2023 core-sales comparison; author analysis.
    4. BEA: What Is the Retail Control Method? Published August 3, 2009; modified September 27, 2024. Extrapolation and reconciliation for the covered PCE aggregate.
    5. BEA NIPA Handbook, Chapter 4: Estimating Methods, December 2024 edition. Retail-control method and illustration, printed pages 4-12 through 4-14.
    6. BEA NIPA Handbook, Chapter 5: Personal Consumption Expenditures, updated December 2024. PCE scope, current estimates, price adjustment and food-services sources.
    7. BEA: Updated Summary of NIPA Methodologies, September 30, 2026. Current source/method summary, including retail control, food services and component deflation.
    8. U.S. Census Bureau: April 2023 Advance Retail and Food Services Release, May 16, 2023. Archived CB23-75 vintage; headline change, sampling margin and March revision.
    9. U.S. Census Bureau: Retail Sales Report and Release Tables. Current release gateway; adjustment language and separate exclusion totals. Content changes with subsequent releases.

    Sources accessed November 14, 2026. Historical release figures retain their stated vintage. Worked examples identified as hypothetical are independently calculated illustrations, not official estimates.

    Disclaimer

    This article is for educational and informational purposes only and does not constitute investment advice. Economic estimates are subject to sampling uncertainty, methodological changes and revisions. Hypothetical examples simplify official methods. Consult current primary releases and a qualified financial professional before making investment decisions.