Why does Baselayer predict the industry instead of using the NAICS code on the state filing?

Because the code on the filing is whatever the registrant typed, and states don't review the submitted industry code.

Secretary of State offices record the NAICS code a business submits at formation without validating it against the business’s actual activity. In practice the filed codes can be generic, outdated, or simply wrong, and the businesses most likely to misreport are the ones you most want to catch, since a fraudulent registrant has every reason to pick an innocuous industry.

Baselayer’s industry prediction is built from what the business actually does: its website, its online presence, reviews and directory listings, its name and address, and its registration details. That produces a six-digit NAICS code with a confidence score you can act on, along with the MCC and SIC codes that map to it. The result reflects observed activity rather than self-reported intent, which is the point of the check.