Skip to main content
Jump to the code table · Choose a data query · Inspect sample records For a product catalog, first decide which records belong in your dataset. A TTB class/type, COLA Cloud category, recorded origin and permittee address answer different questions. None by itself identifies every product on sale. This guide is COLA Cloud’s explanation of data fields. The TTB Public COLA Registry is the official source for approval records; current eCFR Title 27 provides regulatory text. Code descriptions are lookup labels, not complete legal standards.

​
Choose your cohort

product_type is the TTB-oriented product grouping. category and derived_subcategory use COLA Cloud’s consumer category taxonomy; paths can end above a leaf. They are not substitutes for class_name. The API does not document a numeric class-code range parameter—do not send an invented class_from/class_to filter. A permit number links an approval to permit information; it does not establish the current owner of a brand. Sale geography, retail availability and sales require other evidence. Keep your own product ID when several approvals describe a similar product.

​
Try one bounded API query

Choose one example with your existing API key. Each retrieves at most three approval records in the fixed August 2026 date window; it does not download the cohort or promise a particular result. API usage counts against your plan’s shared quotas. These filters follow COLA Search; explicit dates avoid the default recent-history window.
For broader evaluation, see the whiskey dataset, beer data and California wine data pages. Self-service COLA Cloud history starts in 2005; a date filter does not extend the product’s coverage.

​
Three records you can inspect

These are unchanged field values from the 150-row licensed metadata sample, not results promised by the live queries above. Its snapshot is August 20, 2026, with selected approval dates August 13–16. Uppercase CSV column names map to the corresponding lowercase API field names where provided. All three have DOMESTIC_OR_IMPORTED=domestic in the sample. The wine example shows why a product name alone is not a class/type filter. These are published source values, not a legal classification judgment or verified physical-product identity. The sample contains 150 approval records, deliberately balanced at 50 per cohort, with 28 columns. Its whisky_class_100_199 label records the archive selection rule: all 50 selected rows have whiskey class names, but the code-range caveat above matters when applying that rule elsewhere. Only 51 of the 150 rows list a main barcode; missing values mean unknown or not listed, not that a physical product lacks the attribute. The four-day selection cannot measure market coverage. The sample’s CC BY 4.0 grant covers COLA Cloud’s rights in that selection and its derived/published fields; follow its included LICENSE and attribution. It does not grant third-party rights or image access, restrict underlying government records, or license the separate 1,000-row source archive. No images or OCR text are included. See its data dictionary before importing extracted ABV, volume or category fields.

​
Product/Class Type code table

The retained table below is a reference copy with no original capture date recorded in this docs source. It has not been fully revalidated against today’s lookup. For the current lookup, start from TTB’s Public COLA Registry page. The official Public COLA Registry user manual, version 3.11.4 dated November 5, 2015, describes the lookup on printed pages 21–22. Its Figure 15 also shows flavored-rum codes 104–108; that historical example does not establish the current status of every row below.