This section describes how Geometry Search decides which business locations fall inside a supplied boundary, and how to read the confidence score attached to each of them.
How matching works
Every Veridion location has a position, and some also have a building footprint. Which of the two is available determines how a location is tested against your boundary.
Where a building footprint is available, the API measures how much of that footprint falls inside the boundary. A footprint largely inside the boundary is a building on that site. One barely touching the edge is a neighbouring site's building, and is not returned.
Where no footprint is available, the API tests the location's coordinate against the boundary. This is common for sites with no permanent structure: yards, open storage, solar and wind farms, substations, and similar assets.
Footprints and parcel boundaries come from different sourcesVeridion's building footprints and the parcel boundaries you supply are produced independently, so they will not align perfectly. A building genuinely on a site can have a small part of its footprint falling outside the boundary you send.
This is expected, and it is why a footprint does not need to be entirely inside the boundary to be treated as a full match.
How to read the results
Each returned location carries a match_details object explaining how it qualified.
| Field | Meaning |
|---|---|
confidence_score | How confident we are that this business operates inside the boundary, from 0 to 1. |
match_basis | Whether the location was matched on its building footprint or on its coordinate. |
overlap_ratio | For footprint matches, the share of the footprint that falls inside the boundary. |
A score of 0.9 or above means a match we can stand behind: a building substantially inside the boundary, or a position we hold with confidence.
Lower scores mean the match is plausible but not certain. Two situations produce them. A building can genuinely span two sites, constructed across a lot line or sharing walls in a terrace, in which case occupancy is ambiguous. Or the position we hold for a business may be approximate, in which case it falling inside the boundary is weaker evidence than a footprint would be.
Filtering on confidence
min_confidence_scoreapplies the threshold in the request itself, so lower-scoring matches are excluded before the response is returned rather than filtered out afterwards.For occupancy verification, where a wrong attribution is costly,
0.9is a sensible starting point. For discovery, where you would rather see everything and judge for yourself, leave it at the default and read the scores.
Making the requestFor the route, the request body, the query parameters, and the error responses, see Get locations in a geometry.