Tony Lea
Compusearch Micromarketing Data and Systems

Position Statement
Curriculum Vitae
Address

Position Statement

GIS and The Promotion of Geographical Analysis in Business - A Pessimistic View From The Trenches

The following observations and thoughts relate to the use of spatial analysis and GIS in business, especially the marketing and retail site location dimensions of business.  To assist in understanding the context of the remarks, here are a couple of sentences of background information.  I am a former academic who has taught both spatial analysis and GIS (University of Minnesota, University of Toronto, Ryerson Polytechnical University).  I have spent many years (since 1980) working as a consultant in the area of “market analysis” with a strong geographic or spatial component.  I am currently senior vice president at Compusearch Micromarketing Data and Systems which has recently become a division of the Polk Company (Detroit).  Compusearch provides a wide range of data, software, project work and consulting in the area that has become known as Business Geographics.  The majority of our clients are private firms in almost every sector including retailing, financial services, automotive, telecommunications, packaged goods, media, etc.  The data we sell includes a wide range of census and administrative area boundary files, census data, street and highway files, estimates and projections, automotive and track registration data, point files including facilities such as shopping centres, department and mass merchandise stores, grocery stores, etc.  Our premier software products are user friendly business/marketing analysis software systems with all relevant data sets integrated in turn-key systems.  The packages are very geographical by design and feature customized (ESRI) ArcView applications or (ESRI) Map Objects as the geographical interface.

There has been very significant progress in the last 10 years or so in designing and producing user friendly GIS oriented software with huge amounts of diverse data directly accessible to marketing and business users.  Five years ago the most powerful of such systems were “Compass” by Claritas, “Infomark” by Equifax - National Decision Systems and “Conquest” by Donnelly Marketing (all of the U.S.) and “Mosaic Systems” by CCN Marketing (Nottingham, U.K.).  In addition, at that time, thousands of businesses of all kinds used one of the following less expensive (and less integrated) GIS software packages. MapInfo, Atlas GIS, Tactician, Transcad, SPANS, Maptitude, SCAN US and several other small shape packages

The first mentioned 4 integrated packages made geographical analysis and mapping very close to “push-button” and were designed for relatively unsophisticated and non-quantitative users like marketing managers and real estate researchers.  On the other hand, the various general-purpose GIS desktop systems were viewed as ‘lower end’ hands-on type of systems – demanding an in-house analyst (typically a geographer) – by the business community.  Data had to be purchased separately and often took a long time to get set up for easy use by the software.  In general, however, the standard GIS packages were very slow especially on larger (national and state scale) data intensive applications.

Up to about 5 years ago both the integrated geodemographic/GIS packages and the general purpose desktop mapping software systems were used to provide answers to very simple questions and the methodology was almost always very simple.  Some typical applications follow:

1. Geocode customer addresses
2. Map customers in a market as dots
3. Map customers, or customer dollars, over household penetration as small area choropleth maps
4. Create circular trade area around a store and extract absolute and relative profiles of the people and households who live there
5. Create a geodemographic profile of customers based on weighted census variables or membership in geodemographic clusters
6. Create choropleth maps of all the block groups in a market based on extent of matching to the profile in 5

Most of these tasks barely qualify as analysis let alone geographical analysis.

How have things changed in terms of business users objectives, tools and analysis in 1998.  Regrettably, very little in terms of analysis.

First the desktop GIS software field has narrowed considerably.  A new entrant ArcView from ESRI quickly gained substantial market share.  Now MapInfo and ArcView likely have 90-95% of the business market for desktop GIS software, outside of the higher end market for the integrated type of geodemographic/GIS systems (referred to above).  Several of the other software packages have virtually disappeared from businesses – AtlasGIS and SPANS – as the GIS software industry consolidates.  The other small share packages have lost share.  There is now less choice in low-end desktop software. The top dominant packages have added very little new functionality especially in terms of analysis.

There has been a very significant restructuring in the American geodemographics and business GIS industry.  First Strategic Mapping bought Donnelly Marketing and discontinued Conquest.  Then ESRI bought Strategic Mapping, kept the Atlas software line and sold the data business to Claritas.  More recently, last year, Claritas bought Equifax-NDS, its long time arch rival.  It is very likely that Claritas will discontinue the Infomark geodemographic data and software package.  Claritas has reworked its Compass product with a MapInfo-based GIS engine and given it a new name.

There has clearly been a substantial reduction in competition and in choice for users in this industry in the U.S.  The general philosophy of the few big remaining players is to try to focus even more on the off-the-shelf mass market and add only very simple (but usually sexy) new functionality for business users.  The number of buttons in the software has increased but the sophistication has not.

There are likely two exceptions to this general trend and both involve vertical applications.  Lower end sales territory optimization and truck/bus routing applications have become more sophisticated.  Both these applications now permit some interesting analyses at a reasonable price.  But, perhaps the highest demand applications-relating to retail site evaluation and sales forecasting – continue to be overly simplistic – or even crude when evaluated by the standards of spatial analysis.  A few specialized consulting firms offer services to build spatial interaction models market by market for interested financial institutions (especially banks) and large retailers, but in general these approaches are viewed as too expensive and perhaps “over-kill”.

It is very clear that the software and data vendors of this industry view most business GIS users as:

1. Extremely price sensitive
2. Rather unsophisticated (“keep it simple, stupid”)
3. Quite non-analytical
4. Wanting simple answers quickly without much concern for the quality of the numbers
5. Still titillated by colour graphics and maps
6. Unwilling to become involved in new R&D processes

There is a good chance that the software designers and data vendors are right.

I have been involved in putting forward many research proposals to major North American firms to build high quality GIS-based systems for their distinctive business problems.  It is a very hard sell.  In general, business users really do want to keep things simple and avoid esoteric methodological issues.

I have mixed views of how things could trend in the future. I think that it is possible for a group of exciting and competent geographers to get at least some business users excited about doing things right or more rigorously. However, with the number of charlatans around praying on the business people who just want a number quickly, and with the general price sensitivity of businesses, I think it is  really unlikely to happen. On the other hand things could be worse - with simpler software and even cruder “analysis tools”. Business users could demand even lower priced and quicker ‘answers’ to their questions.
There has been some calls in papers in the trade journals for this type of “progress”.

For those of us who are delighted to become involved in genuine spatial analysis to derive the highest quality research for our business and related clients the remarks above have the following implications:
1. High-end spatial analysis research for the business community (with a few exceptions) is done by dedicated researchers without the full knowledge and support of the clients – since they are not being paid for all they are doing for the client.
2. Clients never seem to pay full price for this high end research – much of the real quality analysis is thrown in gratuitously.
3. Most of the analysis work is done in software outside of GIS proper– either in statistical packages or in database packages of in 3 or 4 GL programming languages.
4.  The work that is done in GIS is high end GIS functionality and this seems inevitably to be done (until lately) in ArcInfo.
5. Much time is spend moving data between the various software packages that seem to be best for different purposes
6. After great analyses are completed, analysts have to spend often more time than the analyses took on figuring out how to present the methods and findings as “essentially simple”, to please the executives who write the cheques..

I believe that there are few forces capable of intervening in this state of affairs which will effectively promote higher quality analyses and  models to businesses at least on a scale that is large enough to make a really noticeable dint.

Final Remarks .
 I have asked a number of my colleagues at Compusearch, in academia, and in industry  to make some comments on this draft and to try to prove my apparent cynicism wrong.
I suspect that I will receive some good feedback from this document and would be happy to update this paper with new views in the near future.
 
 
 



 
 

Curriculum Vitae

Curriculum Vitae

Anthony Carson Lea, Ph. D.
Tony Lea is currently the Senior Vice President of Custom Research and Consulting at Compusearch Micromarketing Data and Systems (a division of Polk Canada Marketing Services Incorporated). He received his Ph.D. in Geography from the University of Toronto and has previously taught on the faculties of the University of Minnesota, University of Toronto, Queen’s University, and Ryerson Polytechnical University.  His academic research focused on applied locational models and computer applications, statistical analysis, mathematical programming, and operational research methods applied to spatial problems, all within the framework of geography.  Before and during his period of university teaching, Tony was involved in consulting projects that most frequently involved applied locational research, and the construction of retail site evaluation models for retail chains.  After several consulting projects involving Compusearch, Tony joined the firm full time as Research Director.  Compusearch is regarded as one of the top firms in the world for its innovative and powerful predictive models and optimization models for a wide range of geographical business and marketing oriented problems.  Under his direction, the Custom Research and Consulting Group has built sophisticated site evaluation and other models for retailers and many well known banks in Canada and the United States.  Over the past 20 years, Tony has presented over 60 papers on applied spatial analysis and modeling, and marketing analysis and methods at professional conferences, and has published over 25 papers and books.  He has been a columnist with the industry’s premier magazine Business Geographics.  He is also a member of the Ontario Retail Sector Strategy Advisory Board and was chair of its subcommittee dealing with retail data, research and strategic issues.  He remains an Associate Member of the Graduate Faculty in the Department of Geography at the University of Toronto.


 

Address

Tony Lea,
Senior VP,
Micromarketing Data and Systems
330 Front Street,
Suite 1100,
Toronto, Ontario
M5V 3B7
T (416) 348-9180 ext 429
F (416) 348-9195
Email: tony@polk.ca
Email: tony@polk.ca


Go back to list