GISdevelopment.net ---> GITA 1997 ---> Building & Supporting Application

Non-utility customer information—What’s the value?

Dean A. Zastava
UGC Consulting 6200 S. Syracuse Way, Suite 222 Englewood, Colo. 80111
(303) 773-6166, (303) 773-6618 (fax)


Abstract
Business Geographies applications, data providers, and consultants have created a high level of interest in non-utility customer information due to the increasingly competitive utility marketplace. A utility company today can purchase literally gigabytes of information about non-customers. These data range from digital images that show the location of all buildings to economic, cultural, and education profiles of residents within individual buildings. These data are certainly interesting; however, they are not available without cost, and they may not bring added value to the utility if the data are not analyzed properly, or if the data are purchased for an area that does not need to be analyzed.

Frequently-asked questions address the potential value of non-customer information to a utility company, and what requirements need to be considered before loading this data into an AM/FM/GIS.

Key factors inresolving theseissues willinclude thefollowing:
  • Definingthe need
  • Resolving the address matching problem
  • Determining a relatively low-cost approach
  • Looking into the crystal ball to see what we can expect in the near future
Defining the Need
Several AM/FM/GIS application opportunities exist to further a utility’s goals of adopting a proactive approach to compiling and using non-customer information:
  • Target Marketing - Use non-customer characteristics, estimated new load (inferred age of desired load appliances such as furnaces and water heaters), proximity to existing facilities, and estimated cost to provide service to determine the best geographic areas and potential new customer sites to target for new service. Use targeted analysis results for target marketing promotional campaigns and sales prospect lists.

  • Sales Forecasting - Use new customer growth location areas and non-customer locations and actual or estimated usage patterns (lifestyle profile, age, family makeup) to determine geographic market penetration, forecast geographic market sales and estimate projected new market share.

  • Sales Tool - Predesign distribution for subdivisions to allow “what-if’ costing during, for example, homeowner association meetings. Confirm new applications and contracts while non-customer interest is at its highest point. Reduce follow -up time and avoid missed opportunities due to customers losing interest due to passage of time.

  • Sales Resource Management - Use forecasted sales by geographic market area to determine appropriate sales staffing for each area. Track forecasted-to-actual sales by geographic market area to determine where sales goals are being met and where problem areas exist. Focus sales efforts on areas that are strategic and avoid areas where system is weak.

  • Facilities Expansion Planning - Use forecasted sales revenues and margins by geographic market area to determine the financial payback associated with facilities expansion in each area.

  • Support for Other Utility Opportunities - Gas utility companies are already providing services to municipal water utilities, and all utility companies are looking at meter reading and deciding whether to expand meter reading services to include other utilities in common service areas. Addresses for all buildings (customers of all utilities) are required to meet these new service offerings.

  • Emergency Response - Provide location and other relevant information (e.g., phone number) of non-customers as well as customers for use in emergency response situations.
Many of the applications listed above are strategic and deal with benefits that are softer and more difficult to measure than enhanced labor productivity; therefore, executive support may be imperative to move forward. Be prepared to spend some time educating your executives as to the value of non-customer addresses and information.

Resolving The Address Matching Problem
A top priority for virtually any competitive-minded utility today is to know the location of its customers and potential customers in relation to the location of its facilities. This is not a trivial issue. Utility companies’ customer information systems have historically focused on getting bills delivered and knowing meter locations, and have not been concerned with addresses that were not part of a billing or a service location. Decades of doing business using paper-based maps and records, and relying on cost-plus rate making rather than market forces have left many utilities with a deficient understanding of their market areas and their customers.

To utilize a customer’s or non customer’s location in an AM/FM/GIS, geographic coordinates need to be assigned to each customer and non-customer address. This is accomplished using “geocoding”, the process of matching a data file of address information against a geographically located dataset such as street centerlines (recommended minimum) or Zip+4 centroids (50,000-foot level) sothata latitudellonghude orothercoordinate valuemay be assigned toeachrecord.Thisthenallowstheaddressinformation tobegraphically displayed withinthe AM/FM/GIS.

A recent pilot project for a gas utility client demonstrated that commercially available geocoded addresses from two different vendors was nearly as good as addresses that were digitized by a data conversion contractor from digital aerial photos. The commercial data was deemed adequate from an accuracy perspective, and the cost of all addresses from the commercial data vendor was less than having the data conversion contractor digitize the services. There is also a related problem in obtaining addresses from certain municipalities in the service area without performing a field survey. Progressive data conversion contractors are evaluating commercially available geocoded addresses as a more efficient and lower cost way of providing their clients with address information.

A number of components are required to evaluate the usefulness of such a geocoding strategy. These include a data file of existing utility customer addresses, street address file linked to streets centerlines or Zip+4 centroids, a geographic layer showing utility infrastructure and an address matching and geocoding software program.

Customer Address File
The first piece needed in conducting a geocoding evaluation is a data file with current utility customer address information. These data are acquired by performing an extract of the customer information system to create an ASCII text file containing a number of fields, including:


Street Address File
The street address file uses address ranges on street segments as a reference to interpolate the location of a given address. For example, if a street segment is known to have an address range from 500 to 598 on one side of the street, house number 550 would be located (interpolated) at approximately the midpoint. Street address file information is available from a number of commercial and government sources.

Utilitv Infrastructure Locations
To check the commercially-derived address locations for accuracy, a layer of utility infrastructure locations derived from digital orthophotography and field checks may be used to compare how far apart similarly addressed houses were located between sources. To derive the infrastructure locations, the GIS system may be used to create a view of street centerlines, address numbers, street names, and infrastructure locations. This view then may be exported into a DXF file and imported as a geographic map layer.

Address Matching and Geocoding Software
The ASCII extract of customer address from the utility customer information file is compared to the Zip+4 compliant commercial address data to identify and resolve all mismatches. Most utility companies have performed some level of address matching and cleanup process; however, these processes sometimes rely upon postal workers returning improperly addressed items. This is not a reliable process since the postman is more interested in delivering the mail then in performing an audit of address numbering and street spelling. If the postman recognizes the name of the addressee, the mail gets delivered. If your company has not done a Zip+4 audit of customer address, this is the first step that needs to be done, and you should consider contracting this work out to a company who is expert at it.

The commercial data address data is built from telephone books and other sources and is audited in several ways including compliance to the postal Zip+4 database. In addition, the commercial address data is updated every six months to keep the resident’s name and profile data current. Geocoding (digitizing of address locations), like the Zip+4 audit is best done by company who is expert at it.

Geocoding to be successful, requires tools, processes, and previous experience in gee-coding. Cost when contracted out is on the order of $40 or less per one thousand addresses depending upon the volume. If you are not yet convinced to contract for address matching and geocoding work, read on about some of the issues that must be dealt with.

Locational Accuracy of Geocoded Data
The locational accuracy of geocoded data is dependent upon two factors:
  • Street segment accuracy (horizontal positional accuracy)
  • Address completeness and accuracy (attribute information)
Street segment files with address range attributes were obtained at minimal cost from the U.S. Census Bureau’s TIGER/Line file and evaluated for their positional accuracy. In addition to positional accuracy of the street segment network, the issue of attribute error must be addressed. For example, if erroneous address range information is associated with a street segment (regardless of how position ally accurate the segment may be), any information derived from this dataset will have locational errors. This is due to the interpolation methods used in geocoding, which rely on address ranges along individual street segments. To test the combined accuracies of both the horizontal accuracy of the street segments and the completeness of each segment’s attributes, geocoding results of a utility’s customer address data may be compared to service tap locations as ground truth. Inaccurate and Unmatched Records In geocoding the customer and non-customer datasets, some records may be unable to be assigned accurate geographic coordinates. One or more reasons may exist for these inaccurate geocodes.

1. Non-matching records due to:
  • Missing street segments and address ranges (such as new residential development)
  • Incomplete or missing address information
  • Non-standard U.S. addresses, such as French spellings (e.g., 1110 Rue Bordeaux)
2. Inaccurate locations due to:
  • Erroneous address information (address information linked to wrong street segment)
  • Assumption of even interpolation along a street segment (in reality, addresses are unevenly dispersed along a street)
  • Streets such as cul-de-sacs have many addresses clustered at one end of the road (this example is not represented in the street network file)
3. Multiple addresses found at one location due to:
  • Apartment complexes and multifamily dwellings
  • Office buildings and office parks
  • Retail districts
  • Non-street segment matches matched to a Zip+2 centroid
Determining A Relatively Low-Cost Approach
Premise identifications representing customer locations often proves to be both difficult and time-consuming during an AM/FM/GIS implementation. An alternative methodology is known as geocoding, or locating customers using a commercially available landbase and address file, which matches and plots addresses to road segments. If some 10SS in positional accuracy is tolerable in exchange for cost savings, ease of implementation, and additional attribute information, then this alternative should be given strong consideration. Another potential advantage to using a commercial landbase and geocoding service is that non-customer information can also be easily obtained and mapped into the Iandbase. Including non-customer data can greatly facilitate the use of a number of AM/FM/GIS software applications, especially Sales/Marketing and Emergency Response Outage Analysis.

Adding Non-Customer Information
A variety of issues pertaining to non-customer data should be explored to determine:
  • The feasibility of acquiring commercially available non-customer data for use within the GIS
  • The accuracy and cost associated with using this data
  • Additional marketing strategies that might be developed using non-customer data
To explore the feasibility of acquiring non-customer data, several data providers may be invited to have their products evaluated. Typical data files built by these information providers contain residential, commercial, and industrial locations collected from a variety of sources. For an additional fee, the data files may be geocoded (that is, have latitude and longitude coordinates assigned). Once this process is completed, the data then maybe analyzed for accuracy, completeness, and cost using comparable customer point data derived from address files. This allows an estimation to be made of the expected accuracy, completeness, and cost for non-customer data that might be acquired for other geographic areas both inside and outside the existing utility service territory. In addition to providing location data for non-customers, commercial sources offer a wealth of information regarding the characteristics of specific households and businesses. This information is valuable in segmenting customers to determine service requirements and preferences. Following is a list of data elements available for residential and business sites.

Residential
Name
Telephone number
Home owner or renter
Occupation code (of head(s) of household)
Age of head(s) of household
Marital status of head(s) of household
Length of residence
Income code of head(s) of household
Dwelling type (single or multi-family)

Business
Business name
Key contact names and titles
Primary and secondary SIC codes
Telephone
Number of employees
Year business started
Annual sales

Non-customer Aggregate Level Demographic Information
In addition to site-specific information, commercial sources of non-customer data also supply consumer demographics information representative of typical characteristics for an area or neighborhood. While commercial suppliers attach this information to their records, the most common source of this type of information is through federal government sources. This includes data compiled by the U.S. Bureau of the Census. A census of the U.S. population is formally canvassed every 10 years, and population estimates are projected annually. This information is collected for every family and business in the United States; however, it is geographically aggregated at the national, region, division, state, county, minor civil division, places, census tract, and block group levels. All tabulated information is available at these geographic levels. Additionally, population counts are available at the block level, which is roughly equivalent to a city block.

An additional component of viewing census demographics geographically is addressed by the Census Bureau’s TIGER (Topologically Integrated Geographic Encoding and Referencing) boundary files. Boundary coordinates of states, counties, places, census tracts, block groups, and blocks may be extracted from these digital files and processed for use within an AM/FM/GIS. This allows demographic information to be linked to specific geographic areas. Population characteristics may then be queried geographically to determine a population profile at a specific location, or thematic maps may be prepared. Again, this information is available for the cost of reproduction, but can be found as well on the Internet at no cost. Additional demographic information is available commercially through a variety of providers. These include data on consumer demand, service industry statistics, grocery store and retail statistics, and household statistics. Data are typically packaged by industry or geographic segments and is variably priced. This type of information paired with publicly available data and the customer information system can greatly enhance gee-marketing efforts at the micro level.

This recommended approach should also be coordinated with a clean up of the addresses in the customer information system.

Looking Into The Crystal Ball
Demographic and Boundary File Resources are already available on the Internet. This information will continue to become more available, more accurate, and more affordable. I predict that in the not too-distant future you will be able to purchase digital orthophotos along with intelligent street center lines and up to date addressing from a one-stop “information vendor”. The individual data slices will come from different sources and will be packaged and marketed by the information vendor on the Internet. Listed below are some of the Internet locations where pieces of this data are available today: Also be sure to check out the home pages of the various satellite imagery companies who are aggressively moving into the information provide position in the AM/FM/GIS industry. Identifying the location and characteristics of both customer and non-customer data are critical elements of the AM/FM/GIS database and vital to helping competitive energy service providers achieve the goal of pursuing a proactive approach to marketing their services. Customer data may be readily obtained from existing Customer Information Systems (CIS) and can be readily linked to the AM/FM/GIS. Non-customer data typically is obtained from a commercial vendor and includes information about location of premise, occupancy of premise, demographics of households, and industry, size and other characteristics of business locations. One challenge which arises when moving to full AM/FM/GIS project implementation is to determine how to either obtain and/or geocode customer and non-customer data as accurately and efficiently as possible so that both may be used with other AIWFIWGIS applications.

© GISdevelopment.net. All rights reserved.