Industry classification is the practice of tagging a company with a standardized code that describes its primary business activity. The code answers one question: what does this company do? A staffing agency gets a different code than a software vendor, which gets a different code than a hospital, and any system that reads those codes can segment, route, or count companies without a human reading each website.
That's the theory. In practice, most B2B databases hold a mess of self-reported labels, vendor-specific categories, and free-text guesses, which is why "industry" is one of the least trustworthy fields in the average CRM.
How do NAICS, SIC, and GICS differ?
Three code systems cover most of what a B2B team will ever touch. They were built by different organizations, for different purposes, decades apart, and they don't map cleanly onto each other.
| System | Structure | Maintained by | Where you'll see it |
|---|---|---|---|
| NAICS | Hierarchical industry codes maintained in dated revisions | Statistical agencies in the United States, Canada, and Mexico | Census data, government contracting, SBA size standards, and many B2B data products |
| SIC (1937) | Four digits, frozen since the 1987 revision | No active maintainer; the SEC keeps its own list | SEC filings, OSHA records, D&B and legacy business databases |
| GICS | A hierarchy designed for investment classification | MSCI and S&P Dow Jones Indices | Stock indexes, fund classification, and investor research |
| NACE / ISIC | Four-level hierarchies similar to NAICS | Eurostat (NACE) and the UN (ISIC) | European and international statistics and registries |
NAICS is hierarchical, and that hierarchy is useful. The first two digits give you the sector (54 is Professional, Scientific, and Technical Services), and each added digit narrows it: 5415 is computer systems design, 541511 is custom computer programming. You can roll accounts up to whatever level of granularity your reporting needs.
NAICS is maintained through dated Census revisions, while SIC codes remain present in SEC filing data. Preserve the source and taxonomy version because the same company can be represented differently across systems.
Who assigns these codes?
For NAICS, nobody official. There is no US registry that issues a company its one true code. Businesses self-select a NAICS code when they register with a state, apply for an SBA loan, or bid on federal agreements, and they can pick different codes in different places. A company chasing a government set-aside has an incentive to choose whichever eligible code fits, correctness aside.
Commercial data providers fill the gap by assigning codes themselves. Their taxonomies, update dates, and mapping methods differ, so request the current field dictionary and test a representative account sample before choosing a source. Reconcile profile-selected categories as inputs and verify the final industry code against the chosen taxonomy.
Each source classifies independently. Pull the same company from three vendors and you can get three different answers, especially for firms that straddle categories.
Why do CRM industry fields go bad?
Four failure modes account for most of the damage we see in client databases.
Mixed taxonomies in one field. Your CRM's industry picklist came from somewhere. Then a list import brought LinkedIn categories, an enrichment vendor wrote its own labels, and reps typed values by hand. We cleaned a database last year where "Software," "SaaS," "Computer Software," "Tech," and "Information Technology & Services" were five separate picklist values describing overlapping sets of accounts. Any report grouped by industry was fiction.
Multi-line businesses. Classification systems force one primary code onto companies that do several things. Is Amazon retail, cloud computing, or logistics? The standard answer is to code by largest revenue line, but data vendors don't have revenue-by-segment for private companies, so they guess from the website.
Stale codes. Companies pivot. An agency becomes a product company, a hardware firm moves to services, and the code assigned in 2019 describes a business that no longer exists. Industry values decay slower than contact data, but they do decay, and nobody schedules a refresh for them.
Granularity mismatch. A two-digit NAICS sector is too broad to target ("Manufacturing" spans breweries and semiconductor fabs). A six-digit code is often too narrow for territory design. Teams that rarely decide which level they operate at end up with inconsistent depth across records, which breaks roll-up reporting in a quieter way than outright wrong codes do.
What do B2B teams use industry classification for?
Segmentation is the obvious one. If your product serves logistics companies, you need a reliable way to find each logistics company in your database and exclude everyone else. Beyond that:
- Market sizing. Counting companies per NAICS code against census data is the standard way to estimate how big a segment is. Bad classification quietly inflates or shrinks your addressable market. Our market sizing entry covers this in more depth.
- Lead routing and scoring. Industry is a common input to both. When the field is unreliable, healthcare leads land with the fintech rep and scoring models learn from noise.
- governance and exclusions. Some teams should exclude regulated verticals, government entities, or embargoed categories. That screening runs on industry codes.
- ICP definition. Industry sits alongside employee count and revenue as core firmographic data. An ideal customer profile written against a taxonomy nobody's data matches is a document rather than a filter.
Example
A marketing agency's CRM shows an account named Meridian Health Partners tagged as "Hospital & Health Care," imported from LinkedIn. The company's website shows it's a private equity firm that invests in healthcare businesses. Correct NAICS: 523910 (or 52 at the sector level, Finance and Insurance), and it belongs in the financial services segment, cares about deal flow, and should route to the rep who covers investors. One wrong industry value put it in the wrong campaign, wrong territory, and wrong forecast. Multiply by a few thousand records and the industry field stops being usable for anything.
Common Mistakes
- Trusting a single vendor's label. Each provider classifies with its own logic. Cross-checking against what the company's website says it does catches the misfires, which cluster in exactly the ambiguous companies you most need to get right.
- Letting free text into the industry field. Free entry creates variants that should be reconciled. Lock the picklist, decide your canonical taxonomy, and map imports to it before they load.
- Ignoring the hierarchy. Storing only "Manufacturing" throws away the digits that distinguish a food processor from a chip fab. Store the most specific code you can source and roll up in reporting.
- Classifying once and rarely again. Companies change what they do. A periodic re-check, even annually, keeps the field aligned with reality.
Frequently Asked Questions
What is industry classification?
Industry classification is the practice of tagging companies with a standardized code that describes their primary business activity. The most common systems are NAICS (six-digit codes used across North America), SIC (four-digit codes still needed on SEC filings), and GICS (the taxonomy investors use for public companies).
What is the difference between NAICS and SIC codes?
SIC is the older taxonomy and still appears in SEC filing data. NAICS is maintained through dated Census revisions and is organized hierarchically, so teams should retain the taxonomy name and version with each code.
Are NAICS codes assigned by the government?
No. There is no central registry that issues an official NAICS code to a US business. Companies self-select their code when they register, apply for loans, or bid on government agreements, and different agencies can have different codes on file for the same company. That self-assignment is a major reason industry data disagrees across sources.
How many NAICS codes are there?
The 2022 revision of NAICS defines 1,012 six-digit national industries for the United States, grouped into 20 top-level sectors. The code is hierarchical: the first two digits identify the sector, and each added digit narrows the definition down to the specific industry.
Why is the industry field in my CRM wrong so often?
Because it usually comes from self-reported values, free-text entry, or a vendor's proprietary taxonomy rather than a standard. Reps type whatever seems close, imported lists carry LinkedIn's categories, and enrichment vendors map to their own labels. Layer three sources into one field and you get values like Software, SaaS, and Information Technology describing the same account.
Can Verum standardize industry codes in an existing CRM?
Yes. We review the submitted accounts, propose a consistent industry value, and retain the source, taxonomy, code version, confidence, and exception status with the delivery.
Related Terms
Related: All Glossary Terms | CRM Hygiene | Enrichment Services | Cleaning Services