How a climate-risk analytics provider improved its models with Veridion's facility-level location data.

Registered addresses smooth over real exposure

The customer builds physical climate-risk models (flood, wildfire, heat stress, drought, sea-level rise, hurricane, and convective-storm exposure) for institutional clients across financial services, corporates, and the public sector. The product is delivered as analytics: a portfolio-level view of climate risk, resolved at the asset level so that clients can quantify exposure across the operating sites their portfolios touch.

The data foundation for that analytics layer needs three things from each asset record. It needs coverage: every operating company in a portfolio, with every operating site, not just HQs, because climate hazards hit specific facilities and a multi-site corporate has facility-level exposure that a registered address can't capture. It needs type classification: manufacturing versus distribution versus office versus retail versus R&D versus warehouse, because the model's economic-value attribution and the hazard's relevance vary by facility function. And it needs source-attributed positioning: latitude and longitude precise enough to resolve building-level climate exposure, with the provenance its customer-facing analytics output can defend.

Registry-rooted firmographic feeds (the data foundation most company-data products run on) carry registered addresses. Registered addresses are HQ-level and don't include operating-site granularity. Climate-risk modelling on registered-address data underestimates exposure for any company with more than one site, which is most operating companies above the smallest end of the SMB segment.

400M+ locations, building level, function typed

Veridion's location graph runs to 400M+ locations across the operating-company universe, each carrying source-attributed positioning, building-level precision, and a three-level facility-type taxonomy: 24 L1 categories expanding to 100+ L2/L3 tags. The taxonomy distinguishes manufacturing facilities from distribution centres from offices from retail outlets from R&D sites from warehouses, with sub-classifications underneath each, at the level of granularity climate-risk models need to make function-aware exposure attribution.

The customer integrates the location graph as a structured data layer underneath its modelling pipeline. Every operating company in a client's portfolio resolves to its full set of operating sites in the graph, each typed by function, each positioned to building-level latitude/longitude with confidence and source attribution. The model runs against that asset-level resolution rather than against HQ-level proxies, and the analytics output that lands with its clients carries the same per-attribute evidence chain the location graph itself provides.

Risk resolves to the asset, not the proxy

For the customer's clients, the practical effect is that physical climate-risk analytics resolve to the assets that are actually exposed, building by building, function by function, rather than to a registered-address proxy that smooths over the very thing the model is meant to surface.