Kind of Business

Kind of Business is how Census retail statistics name their rows. Each entry couples a NAICS code with the phrase shoppers would recognize - 'Family clothing stores', 'New car dealers', 'Automotive parts, acc., and tire stores' - so a workbook reads as a table of retail lines rather than bare codes.

In the US Census Monthly Retail Trade Survey (MRTS) dataset used for apparel retail, the field sits as Column B of every yearly sheet. The motor vehicle and parts dealers cut of MRTS carries the same label field, which is where 'New car dealers' and 'Automotive parts, acc., and tire stores' appear as distinct rows.

The rest of the sheet is calendar: monthly columns in millions of dollars extend across the year, and two marker values keep the numbers honest. '(S)' flags estimates withheld to avoid disclosing individual firms, and '(NA)' marks cells Census does not publish at all.

The label column decides whether a retail dataset can be joined to your own taxonomy. If your product categories are 'clothing' and 'auto parts', the MRTS labels tell you instantly which rows to aggregate - 'Family clothing stores' versus 'Automotive parts, acc., and tire stores' - without reverse-engineering six-digit NAICS codes.

It also governs how you handle missingness. Because '(S)' marks suppressed-but-real estimates while '(NA)' marks unavailable ones, a parser that treats both as blank will silently undercount apparel-retail sales in months where disclosure rules bite. Retail analysts comparing homefurnishing-retail trends against automotive-retail ones need those distinctions intact, which is why the label-and-marker convention matters more than the raw dollar figures.

1. Check the label-to-code pairing. In MRTS yearly sheets Column B pairs each label with its NAICS code; confirm both survive your ingest so 'New car dealers' stays distinguishable from other transportation-equipment lines.

1. Verify suppression markers survive parsing. '(S)' and '(NA)' must remain distinct values; collapsing them erases the difference between withheld and nonexistent.

1. Confirm units travel with the column. MRTS monthly figures are in millions of dollars - a loader assuming dollars inflates every retail line by a factor of a million.

1. Test one sheet per year. The apparel-retail MRTS extract repeats the Kind of Business structure on each annual sheet, so parse the first two sheets before trusting the rest.

1. Match labels to your category tree once, centrally. Map 'Family clothing stores', furniture and tire-store labels into your taxonomy in one place rather than per report.

Entries that pair with Kind of Business in this glossary:

• monthly-retail-trade-survey - the survey whose workbooks carry the column.
• naics-codes - the classification codes each label is paired with.