Utilicorp's fame delivers its first marketing application
Mary Ann Stewart P.E. UtiliCorp United 20 W. 9th Street Kansas City, MO 64105 Utilicorp's fame At the end of 1999, UtiliCorp completed an aggressive three year reengineering and conversion project for its mapping systems, taking paper and electronic maps of gas and electric facilities distribution system in eight states to a common electronic format. This reengineering effort focused on moving mapping functions for facilities design from centralized drafting personnel to decentralized construction coordinators in field offices. Applications development for design work based on customized tools proceeded in parallel with data conversion and hardware/network rollout. The resulting AM/FM/GIS of UtiliCorp has evolved to become FAME (Facilities Management and Mapping Enabler). One of the goals of FAME was to provide enterprise-wide pipes and wires information to managers for use as an analysis and decision making tool. The UtiliCorp marketing research department proved to be receptive to an early exploration of these new capabilities. FAME provides very attractive data for marketing research because facilities designers are constantly making entries of their proposed and as-built designs into the main data store. This facilities information can be immediately viewed and queried by the marketing FAME user, and can be combined with other geographic and database information for analysis. Utilicorp energy delivery marketing UED (UtiliCorp Energy Delivery), the "wires and pipes" division of UtiliCorp United, consists of the transmission and distribution network. These regulated operations serve franchised territories in eight states, with rates set by state or local regulatory bodies. The energy delivery business unit serves 373,000 electric distribution customers in Missouri, Kansas, Colorado and West Virginia and 851,000 natural gas distribution customers in Missouri, Kansas, Colorado, Nebraska, Iowa, Michigan, Minnesota, and West Virginia. UED's organization is centralized around the major functional processes of transmission services, network services, community services, market services, and customer services. Earnings are enhanced through a focus on management of assets, introduction of new products and services and providing customer-focused solutions. Support for growth initiatives comes from the packaging of processes for transfer to new operations, sharing and leveraging of knowledge gained from global delivery experience, and customerbased programs to retain existing customers and expand the network. The retail market for energy is changing. While policy makers debate the nature, extent and pace of change, there is little doubt of one important end result: retail energy customers will have more and different opportunities to shop for electricity, natural gas and energy services in the future. To prepare for this future, marketing objectives focus on increasing services to existing customers, new customer acquisition, retaining existing customers and improving knowledge of customers. In addition, programs are offered to customers through direct sales, telemarketing, and direct mail. There is a substantial appliance service business, offering a tiered system of service contracts, and including such products and services as bill insurance, stand-by generators, duct sealing and cleaning. What is marketing research? UED's Director of Market Research was the internal client for this project. In order to better serve the mapping and analysis needs of this group, it was helpful to understand the group's overall mission. The American Marketing Association, in an article from "AMA News" provides this concise statement: "Marketing research is the function which links the consumer, customer, and public to the marketer through information used to:
At the conclusion of all research processes is the dissemination of information including:
Improving market research processes with spatial analysis Examination of UED's traditional market research analysis revealed a number of cumbersome or limiting procedures. Prior to the development of enterprise-wide mapping capabilities, collecting information for a study was a complex gathering task. Field office personnel and a few centralized drafting pools maintained facilities records. This information was customarily distributed through annual map book updates and by custom mapping projects for an area. Corporate marketing staff would make direct contact with local field personnel to determine the current status of the distribution system for each area of concern. Information was often transferred by sketching on the back of an envelope. The geocoding process, producing latitude and longitude points from addressing information, was being outsourced. This required careful formatting of customer information extracts, sending an extract, waiting for results, and receiving the product of an unexplored black box process. If there were lots of successful matches it was good fortune; if there were lots of misses it was a mystery. Additionally, creation of a mailing list certified for bulk mailing with associated phone numbers was being outsourced. With the arrival of spatial analysis capabilities through FAME, marketing became interested in acquiring more geographically specific demographic information, preferably at the household level. Information about housing stock (age and value of residential housing) was particularly difficult and costly to obtain. We evaluated marketing needs related to FAME capabilities and produced some initial solutions. Facility information for all UtiliCorp service areas would be available as real time geographic and database information through the enterprise-wide rollout of FAME. Facilities designers enter their designs directly into FAME, with subsequent posting to as-built status. Everyone in the enterprise could have immediate, detailed real-time information of facilities at any location. Further, it was possible to compose on-the-fly queries of the facilities information. We determined that it would be preferable to bring geocoding capabilities into FAME in order to streamline and better understand the elements of this process. We evaluated various geocoding tools for use in FAME and ended up using a very easily implemented stand alone tool. This geocoder provides current addressing information for the entire United States, has a good interactive interface for attempting to resolve misses, and provides straightforward latitude, longitude and census boundary information for import into FAME. Conversely, our initial evaluation determined that mailing address and telephone information would best be left as an outsourced service. The search for household level demographic information remained challenging. We made a breakthrough in one service area where we have a geographic data sharing agreement with the city. We were able to obtain the county assessor database which shows current housing values as well as address information for each plat. Northern Minnesota-Sales target areas In general, UED Marketing programs target on new construction, but there is also focus on residential conversion, targeting rural propane customers and urban electric customers with high summer cooling bills. Our first pilot was a typical residential conversion project for gas customers in northern Minnesota. Marketing staff needed to create and prioritize sales target areas in northern Minnesota. They obtained an extract of selected towns from the UtiliCorp customer information system. The extract included customer addresses as well as historic load information. The plan was to plot customer locations with respect to gas facilities. Noncustomer addresses could be derived by an address culling process, as FAME includes address ranges in its landbase. These noncustomers in the service area would be targeted for residential conversion from propane to UED natural gas services. The extract was passed to FAME for geocoding of addresses, resulting in individual customer points with associated database information. Marketing researchers were then able to use FAME to see noncustomer locations adjacent to our gas mains. The goal of the northern Minnesota project was to target fully built out older neighborhoods with many holes (noncustomers). Researchers could use map graphics and generated statistics to prioritize areas to be worked. This methodology literally replaces the technique of staff driving around looking for propane tanks in residence yards. Additional demographic information could then be provided for analysis. Block group profiles were built using standard attributes such as age, income, presence of children in the household. Using block group boundaries in FAME, it became possible to perform spatial analysis on these block group profiles. The products resulting from the northern Minnesota pilot were: 1) an address/phone list of potential customers; and 2) results from graphic analysis and query which could be used to determine future target areas. This pilot was somewhat unique in business geographics applications in desiring household level information rather than aggregated data. This desire may be driven by the availability of a large customer database creating the expectation that a similar level of detail could be obtained for noncustomers (who are the individuals of interest in this targeted marketing). More typical demographic applications rely on commercial products providing annual updates of aggregated census data. There remains a challenge to obtain household specific information when the households lie outside the enterprise's information systems. The second pilot, Kansas Public Service, provided one approach to solving this problem. Kansas public service --Linking county assessor data This pilot project was concerned with making a marketing offer to high income customers in one service area. Various manual approaches had been attempted, including the always popular technique, driving around looking for big houses, in this case. A somewhat more automated approach was to target a zip code believed to be high income. However, the community had only five zip codes, resulting in too large a segment of the population being represented by the chosen zip code. Could FAME be used to determine incomes within the chosen zip code? Fortunately, this service area was one of seven county/municipality entities providing electronic landbase to FAME. We had an electronic data sharing agreement with the city based on cost sharing for periodic aerial surveys. This agreement allowed UtiliCorp access to the county's electronic version of the assessor database, with the understanding that the database was not to be used to solicit new utility service customers. The database was provided in a standard format and could be queried and manipulated using standard query tools. We sorted the database by assessed value, evaluated the range of property values, culled the list to include single unit urban residential properties, and finally had a list of properties to geocode using the address field of the database. Customers were determined by matching latitude/longitude from geocoding results of the UED customer database to the latitude/longitude from the county database, thus employing the address parsing capabilities of the geocoder. We retained the property value information from the link to the county assessor's database, making it possible for the market researchers to see customer information with property value appended on any of the properties. The product resulting from the KPS pilot was a customer list for a targeted marketing offer. Customer names and addresses were sent out to a commercial service that provides phone numbers and certified addresses for post office bulk mailing. The pilot does not provide much in the way of classic spatial analysis. Rather, it demonstrates the use of linking an AM/FM/GIS system to traditional external databases with geographic hooks. County assessor and other government databases, when available, can provide highly specific, current, geographically referenced information. Government sources are significantly different from commercial demographic data sources, both in the information provided and in the business arrangements concerning the data. A geographically large and diverse enterprise may find it overwhelming to obtain local level data from many sources, but should be aware of the possible benefits to be obtained from existing or desired data partnerships with governmental agencies. Problems and interesting issues Some of the pilot problems were easily predicted. There were problems with addresses. The FAME landbase includes address ranges, but only at the level of unenhanced TIGER addresses. The stand alone geocoder helps deal with this issue. It uses a current extract from a vendor-supplied landbase with enhanced address ranges, producing a higher percentage of hits than would be obtained otherwise. An embedded geocoder would search the less satisfactory FAME address ranges. However, there were still a fair number of misses from geocoding. This could be attributed to: bad landbase information, bad customer address information, user error. We found geocoding success to be highly variable depending on area. As expected, urban areas with rapid growth had more timely landbase information with a resultant higher match rate. As discussed earlier, individual household information is difficult to obtain in many cases. The Census Bureau cannot release individual household information due to confidentiality issues, and thus this richly detailed demographic data is always aggregated. Some commercial demographic data providers conduct extensive surveys and perform statistical analysis to extend the survey data to the total population. Our market researchers have not been able to find a satisfactory commercial source for the desired housing stock information. Local government data sources as used in the KPS pilot appear to be a promising solution to the aggregated/individual household dilemma. We were startled by the magnitude of one of the problems. Incomplete facilities information in some locations made analysis difficult if not impossible. In performing the northern Minnesota spatial analysis we found customer locations beyond the range of gas mains. Examination of existing facilities information leads us to believe that recent line extensions have not been entered into the system. Issues of completeness, currency and accuracy of facilities information will continue to trouble FAME during the period of transition from centralized drafting to local data entry responsibility. Our solution to this dilemma in northern Minnesota was-move on to Michigan. Seriously, it is important to understand that the acquisition of relatively homogeneous commercial national demographic and landbase data will in no way produce consistent quality in a utility's facility information. In our case, Marketing's problem with the facilities data has helped move us toward a somewhat painful cleanup of data in northern Minnesota. The future of market research projects and spatial analysis FAME as a full blown facilities design application requires high end hardware, network bandwidth, expensive software licenses, and extensive training. Thus, we have been quite interested in exploring web based applications for marketing view and query functions. Our marketing pilot used one full blown FAME workstation but this has not necessarily convinced us that this is the best way to serve more marketing staff. We currently run an intranet view and query version of FAME and anticipate expanding its functionality for market research. Market researchers are getting a feel for the new kinds of questions that FAME capabilities allow them to answer. They have recently developed an extensive database that pulls information from our customer information system. They have traditionally worked with tabular demographic information and plan to use FAME for spatial analysis of this data. Marketing would like to geographically profile areas and customers, to study what has been profitable historically and to extend this success to similarly profiled areas. Geographic profiling would also be helpful in the targeting and media planning stages of developing a new service area. Routing applications are an intriguing component of FAME spatial analysis capabilities. Traditional routing applications assist in dispatching trucks. Marketers traditionally communicate with potential customers by mail or telephone. However, services offered might well contain a routing component. Questions which could be asked include: Is it practical to extend appliance services to a larger area and, if so, where? What routing synergies exist relative to parts storehouses and other centralized distribution sources? Are there operational efficiencies which could be achieved by locating services in a particular area? UtiliCorp's marketing researchers have worked with FAME and have become staunch advocates of seeing is believing. Over and over, they mention the geographic component of their work in explaining their existing processes. "It would be nice to actually see it on a map" is becoming a common refrain. This seeing process results in deeper, clearer understanding and better communication with others in the organization. | ||
|
|