Polygons are the preferred geometry type for reference data that represents administrative areas, postal codes, and points of interest, because they provide more accurate reverse geocoding results and allow the Create Locator tool to restore missing attributes through spatial linking in multirole locators. Using polygons to build a locator based on the Point Address or Parcel role will optimize reverse geocoding results. On the other hand, using point features as reference data to build the locator reduces the size of the locator, because less disk space is required than for storing polygon features. See Primary locator roles for details on which roles suggest using polygon features for reference data.
For reverse geocoding, if there are no streets near the input location, large polygonal features that represent parks, universities, zoos, or airports may be returned. If the location isn't within the boundary of this type of feature, a postal code or administrative area (such as a city) is returned. If the input location intersects multiple boundaries, the feature with the smallest area is returned. The reverse geocoding search tolerance is limited by a hierarchy of search tolerances or 500m for single role locator or when only a single feature type is selected in the features to support reverse geocoding setting in the Locator Properties dialog box or checked as a Feature Type value in the Reverse Geocode tool. When you are reverse geocoding and want to return results for POI, Parcel, Postal, and Administrative areas, use a polygon to represent these features when building a locator. A reverse result is returned if the input intersects the boundary of the feature. For more information about the hierarchy that is used for reverse geocoding, see the summarized hierarchy table.