Industry codes make data sortable. NAICS and SIC let you segment a list, route a lead to the right rep, or filter a market. The problem is getting them: buying a database is expensive, and labeling by hand does not scale.
The classification call reads a company’s website and returns NAICS and SIC codes for what it does. You pass a domain.
How it works
We read the site, understand what the business sells and to whom, and map that to the standard code lists. It works from the page content, which means it works for the long tail of small businesses that have a website but no clean database record.
Be realistic about it
This is classification from a website, not a registry lookup. For a clear business with a decent site, the result is usually what a human would pick. For a site with a single splash page, or a company that does five unrelated things, expect to review. Use the codes as a strong default in your pipeline, with a human check where the stakes are high.
Where it earns its keep
- Lead enrichment. Add an industry code, a logo and brand colors to every record from just a domain.
- CRM hygiene. Backfill missing industry fields on an old list.
- Market maps. Take a list of domains and bucket them by industry.
Combine it with brand extraction and you get a company profile built entirely from the website. Start on the Free plan.
What goes in and what comes out
You pass a domain and get industry codes back: NAICS, the North American Industry Classification System, and SIC, the older Standard Industrial Classification. Both are widely used in CRMs, lead lists and market reports, which is why having both is handy. A tool that expects NAICS can use the NAICS code, and a legacy export that still speaks SIC can use the other. The result is a strong suggestion drawn from what the site says the business does, not a legal declaration of the company’s classification.
Building a review step that is worth the effort
The honest way to use classification at scale is to separate the easy cases from the hard ones. Run your list, then route the results. Clear cases with a decent site go straight into the record. Anything unusual goes to a person: a company with a thin one-page site, a business that clearly does several unrelated things, a domain that is a parked page. You do not need to review everything. You need a rule for what gets reviewed.
A simple way to start is to sample. Take a hundred domains, classify them, and have someone check the codes by hand. If the agreement is high for your kind of company, trust the output more. If it is low for a certain segment, say holding companies or agencies, handle that segment separately. The point is to measure on your own data rather than assume.
Combining it with the other calls
Classification is most useful as one field among several. A single domain can give you a logo and a palette from brand extraction, the text of the home page from scrape, and an industry code from this call. Put those together and a bare domain becomes a record with a name to display, a look to match and a segment to filter on.
Cost and quota
Classification works against your monthly plan like the other calls. The Free plan is enough to try it on a real sample before you decide whether to run it on a whole list. If you run it across a large CRM, plan the quota first and run it in batches so a mistake in your input does not cost you a month of pages.