Reading the Census monthly retail report alongside POS data

The Census Bureau's monthly retail report estimates the dollar value of sales at US retail and food services businesses, based on a survey of firms. It's one of the most widely cited measures of consumer spending. Reading the census retail report alongside point-of-sale scan data can add useful context, as long as you're clear about what each one measures.

They answer different questions. Knowing the differences keeps you from comparing figures that were never meant to line up.

What does the Census report measure?

The Bureau runs the Monthly Retail Trade Survey, along with an advance version that provides an early estimate shortly after each month ends. Both estimate sales by kind of business, such as food and beverage stores, general merchandise stores, and nonstore retailers.

A few features matter for interpretation. The estimates come from a sample of firms, so they carry sampling error. They're reported in dollars, not units, and they're not adjusted for price changes. They're published both seasonally adjusted and unadjusted. And they're revised: the advance estimate is updated in later releases, and annual revisions can change past months.

The Census Bureau's retail pages describe the methodology, the categories and the release schedule in detail.

How is it different from POS scan data?

In source, first. The Census figures are survey estimates reported by businesses. Scan data is a direct record of transactions at the register, aggregated from the stores that provide it.

In scope, second. The Census report covers retail broadly, including many store types and nonstore sales. Scan data from a particular network covers the stores in that network. For NRS Insights, that means the NRS network of independent retailers.

In detail, third. The Census report classifies sales by kind of business. Scan data can go down to the product, by UPC, by store and by day. It can show units as well as dollars, and it can separate price from volume.

In timing, fourth. Each source has its own release schedule and revision pattern. Comparing an early Census estimate with finalized scan data mixes a preliminary figure with a settled one.

How should you read them side by side?

As context for each other. If both show softening in a similar kind of business over the same period, that's a consistent picture. If they diverge, the divergence is a question to investigate, not an error to resolve.

Match the basis. Compare unadjusted Census figures with raw scan data, or seasonally adjusted with seasonally adjusted. Compare dollars with dollars. Line up the same months.

Keep in mind that the Census figures aren't adjusted for price. If scan data shows dollar sales rising while units fall, you're seeing price increases at work. The Census dollar figures would include that price effect too, without separating it.

What are the common pitfalls?

Treating one as a check on the other. They measure different things. A difference between them doesn't mean either is wrong.

Quoting the advance estimate as final. It's an early estimate and will be revised. If you cite it, note that it's preliminary.

Extrapolating from one to the other. Independent-store scan data describes independent stores. The Census report describes retail broadly. Neither should be scaled to stand in for the other. The NRS Insights monthly reports describe the independent-store channel, and the July 2026 same-store sales report is the most recent.