Resolution methodology

Resolving an address means answering two questions in sequence: where is this place, and what business operates there.

The API is designed to work with input as it actually arrives: free text, inconsistently formatted, sometimes incomplete, and often pointing at a multi-tenant site rather than at a specific unit. It leverages several mechanisms to resolve such input to the right business location, and it reports how it got there so you can judge the result.

Locating the address

The input address is first resolved to a geographic point, and classified by how precisely it lands:

All three are accepted. Inputs that resolve only to a broad area are rejected, because they cannot identify a place of business. See How the Location Resolver works for the full list of accepted and rejected input.

Sizing the search area

A resolved point rarely sits exactly on the business you are looking for. Addresses land at building entrances, at the centre of a street, at the gate of a site rather than the facility inside it. So before looking for anything, the API decides how far around the point it is willing to search.

That distance is not fixed. It adapts to how built up the surrounding area is. Rather than applying the same number everywhere, the API widens the search until it has enough nearby candidates to work with, then stops. On a dense city block that happens within tens of meters. On the edge of an industrial estate, or in open countryside, the same search may reach several hundred meters, because there is simply less around.

The point of this is consistency. A fixed distance would be far too wide downtown, where dozens of unrelated businesses sit within a hundred meters, and far too narrow in a rural area, where the nearest business may be the only one for half a kilometer. Adapting the distance means the API considers a comparable set of possibilities wherever the address happens to be.

A wider area is used to find a company you name. When you supply a company name or ID, the API searches a larger area for it than it would use for proximity matching. This is safe precisely because you named the business: a name or ID match is unambiguous no matter how many other businesses were nearby, so reaching further costs nothing in precision and considerably improves the odds of finding the right site when the address resolves slightly off, to a campus entrance, the wrong side of a large facility, or a neighboring street.

That wider area is used only for finding the company you named. If it is not found, the API falls back to the same tighter area it would have used for the address on its own.

Only buildings with a known business presence are ever considered. Purely residential structures, and buildings with no recorded operational presence, are never treated as candidates and are never returned as matches. In a mostly residential suburb the search will therefore reach further than the density of buildings alone would suggest.

Finding the business

When you supply a company

A named company is the strongest signal available, so the API looks for it before anything else.

A confident match returns that business, along with the building it occupies if it has one. Names are matched on how closely they resemble the name on record, so exact names are strongest, close variants score slightly lower, and a name too far from the name on record is treated as no match at all. A Veridion company ID is matched exactly.

Because this search runs first, a confident match can win over the place that proximity alone would have selected. If you name a business and that business is demonstrably at the address, that is a better answer than the one proximity rules would have produced on their own.

A matched company is still described spatially, in exactly the same terms as any other result: inside a building if the point falls within the one it occupies, near a building if the point falls outside it, or as a business location on its own if there is no building. A company match does not imply the point landed inside anything, and a confirmed company can legitimately sit some distance from the resolved point.

If the company is not found, the API does not stop and does not report the address as unresolved. It continues to proximity matching, and reports that a business was found at the address but is not the one you asked for.

When no company is supplied, or the company was not found

The API matches the point to a place by proximity. Three cases apply, in order:

A building is always preferred to a business location on its own. If both fall within range and the business location is closer, the building still wins, because a footprint is a verifiable outline of a place while a coordinate on its own is a single point that may sit anywhere on or near the site.

The third case covers places that have no mapped structure: solar and wind farms, remote or rural assets, and sites where no footprint is available. These return a location with no building attached.

When no business is found

If no business location and no shared building can be identified, the API performs a final check against Veridion's legal entity data. This check exists only to explain why an otherwise valid address returned no business. It never turns a registered address into a location, and it never returns legal entity records.

If the address is recognized as a company's registered address, the API reports this explicitly. This tells you the address exists in the legal universe, but Veridion found no evidence of business activity there. Registered and mailing addresses are never returned as business locations.

If nothing matches at all, the address is returned as unresolved. The API does not return a low confidence substitute, a nearby unrelated business, or a broad area guess.

Behavior summary