Geocoding eludes easy answers, rewards diligence

Geospatial analysis, once the exclusive domain of GIS departments, is fast becoming an integral feature of line-of-business applications in many industries, especially insurance, financial services and telecommunications. Companies want to make decisions, sometimes automatically, based on the characteristics of objects or people located at specific locations.

Geocoding is the process of defining those locations precisely as latitude/longitude coordinates. On the surface, there doesn’t appear to be much difficulty in matching geographical coordinates to a given street address or location name. In practice, however, obtaining reliable geocodes at the pace businesses need them has proven to be an elusive goal. Location Intelligence experts at Pitney Bowes have heard from numerous developers about their quests to find enterprise-level geocoding technologies. Disillusioned with free geocoders and frustrated with the inflexibility of some off-the-shelf software, developers are seeking a better approach to procuring a geocoding solution.

Based on these interactions, Pitney Bowes has published an eBook designed to assist in the procurement process. Geocoding for Business lays out five key criteria for selecting a geocoder:

  1. Matching, the interpretation of an input address and matching it to reference data
  2. Positional accuracy, how close the geocode is to the target addressable location
  3. Throughput, how many records per unit of time the geocoder can process
  4. Metadata, the data about the geocoding results, which enables operational automation and insights
  5. Deployment options, such as operating platforms, on-premises or cloud hosting and integration with existing applications.

The five criteria and several illustrative use cases form a concise, easy read, serving as a useful first step in compiling a geocoding solution shortlist.

The eBook defines match rate as the proportion of input addresses that the geocoder accurately matches with the reference streets, boundaries, addresses and so on. The higher the match rate, the better, but there is a risk of false positives, which can pose serious problems if the geocodes serve as the basis for automated decision-making processes.

Another potential pitfall is the misunderstanding of positional accuracy. It may seem that the ideal geocoder is the one that identifies the front door of every target address. And that may be necessary for some applications, such as insurance underwriting. But in other use cases, such as demographic profiling, neighborhoods or street centroid may be enough. Flexibility is key: The geocoder needs to be able to accommodate the varying use cases that may present themselves within a single business.

In its modern form as an embedded application component, geocoding needs to be fast if it is to avoid becoming a bottleneck. Pitney Bowes pegs the acceptable throughput at hundreds of millions of records per hour. It’s not just new queries that fuel the need for such speed. Enterprises such as insurance companies update geocodes for their entire books of business every month, a feat that is only practical with a high-throughput geocoder.

Other guidance for enterprise-level geocoding, such as application integration and operating environments, appears at the end of the eBook.

According to the Pitney Bowes experts, metadata (data about the geocode) is a key enabler of the next generation of spatial analysis tools, broadening the scope of insight on a target location and its surroundings. For example, metadata about adjacent office suites within the same building as a target office or demographics of the patrons of neighboring businesses can provide a significant competitive edge in business decision-making.

To learn more about the selection criteria for enterprise-ready geocoders, download this eBook by Pitney Bowes: Geocoding for Business: Considerations for selecting geocoding technology to support a winning Location Intelligence business strategy