Track NAICS versions without blindly replacing codes

Published by Federal Bid Partners LLC · October 26, 2026 · Versions

A code number without a version can be ambiguous in a research file. Classification systems change, and historical records may use earlier definitions. When comparing business profiles, award data, or old research notes, record the classification year before concluding that two codes mean the same thing.

The Census NAICS site provides manuals, reference files, and concordances. As reviewed on October 26, 2026, it presents the 2022 search and notices concerning proposed 2027 updates. A proposal for a future revision is not proof that a new code is already usable in a particular system.

Identify the source and version

Write down the system, publication or data vintage, code, title, and definition source. A current website, a historical award record, and an old spreadsheet may each need a different interpretation. Preserve the original value in historical data rather than overwriting it with a current guess.

The BLS classification overview explains where Census concordances can be found. Use a concordance to identify relationships that need examination. Check definitions when a relationship is not straightforward; a crosswalk is not a substitute for understanding the work.

Fictional example: an old customer list

Fictional Orchard Research inherits a prospect list built under an earlier NAICS version. Its analyst wants to replace every code using a spreadsheet lookup. A review discovers that some old categories do not translate cleanly into one current category.

The analyst keeps the original column, adds the source version, and creates a separate candidate-current-code column. Ambiguous rows receive a manual review flag and an explanation. The team can now reproduce the transformation without pretending every mapped row is confirmed.

Respect the system's adoption rules

FAR 19.102 states that new NAICS codes are not available for federal contracting until SBA publishes corresponding size standards. Check the solicitation and relevant official guidance before changing procurement records. Do not assume a Census revision announcement automatically updates every federal system.

For recurring research, save the transformation method as well as the result. Note which concordance was used, which direction it maps, what happened to ambiguous records, and whether totals changed after regrouping. Keep unmapped records visible instead of silently discarding them.

• Record classification year and source system.
• Preserve original historical codes.
• Use the official concordance in the correct direction.
• Review ambiguous mappings against definitions.
• Check adoption and size-standard requirements.
• Document manual decisions and unresolved rows.

Finish with a versioned research file and a small exception log. That is more useful than a perfectly filled column that conceals uncertainty. When a future revision arrives, the same process will let you assess its effect without losing the meaning of the records you already rely on. Before applying a bulk transformation, test several ordinary rows and several exceptions. Ask a second reviewer to reproduce the mapping from your saved instructions and sources.