How a commercial carrier quadrupled its SMB match rate by replacing registry classification with Veridion.

Half of SMBs sit on the wrong industry code

The customer is a Canadian property-and-casualty carrier writing commercial insurance across small-business, mid-market, and large-account segments. Its dedicated SMB-commercial brand is the entry point for the small-business book the engagement specifically supports. Like every commercial carrier writing at SMB volume, the customer runs into a structural data problem at the underwriting front line: the registry data underwriters rely on for risk classification is wrong on the segment that matters most.

The industry numbers frame the gap. More than half of SMBs are misclassified in government registries. Underwriters spend roughly 14 hours a week (about 35% of working time) on tasks that could be automated, most of them tied to validation and classification. Only 5% of underwriters say technology has actually automated their non-core work. For a carrier writing SMB volume through a dedicated brand, that's a structural cap on quote velocity, on underwriting accuracy, and on pricing discipline.

The brief was to put a data layer in front of the SMB underwriting workflow that classified every applicant correctly, returned a complete attribute record fast enough to keep the quote moving, and refreshed often enough that the data underneath today's risk decisions matched today's reality.

Full taxonomy stack at the underwriting front door

Veridion's Match & Enrich API now sits at the front of the customer's SMB underwriting flow. When an applicant arrives, through the SMB brand or via broker channel, Match & Enrich resolves the business name and address against Veridion's company knowledge graph in approximately 1.5 seconds, returning more than 50 structured data points per applicant.

Classification runs across the full taxonomy stack a Canadian commercial-insurance application needs: NAICS, SIC, ISIC, NACE, NCCI, IBC, plus Veridion's own proprietary business tags. Each applicant comes back coded across all of them, with confidence scores and source-evidence snippets attached for auditability. The attribute payload spans business activities at granular taxonomic levels, products and services, multi-location footprint with facility-type classification, digital-presence signals, and risk-keyword scanning against the business's own digital footprint (material that registry-only providers don't carry).

The graph underneath refreshes weekly across 134M+ operating companies, with daily refresh on volatile attributes. Records that change between refreshes (new ownership, new locations, regulatory-status shifts, dormancy signals) flow into the customer's pipeline through the same API, so the data front-end of the underwriting workflow stays aligned with the world the policies are getting written against.

Classification stops being a 14-hour weekly tax

The match-rate lift is the headline number, and the mispricing recalibration is its second-order consequence: when one in five SMB policies turns out to be priced against the wrong industry code, recalibrating on the right code corrects premium leakage in both directions: underpriced risk on businesses that had been undercoded, and overpriced risk on businesses that had been overcoded. The first effect protects the carrier's loss ratio; the second protects the customer relationship.

For the customer's underwriters, the upstream consequence is that classification and validation stop being a 14-hour-a-week tax on the desk. The data arrives correct, fast, and current, and the underwriter's time goes back to the actual risk decision.