GeOdEmOgRaPhIcS*

Click here to start

Table of Contents

GeOdEmOgRaPhIcS*

Geodemographics- A view from 20,000 feet!

Contents (in case you want to leave)!

What does the word mean?

What does the word mean?

The hope is that...

How does it work?

A number of standard systems are widely used in the UK

Building a Geodemographic System

None of the Commercial Systems have ever been subject to detailed scrutiny by external and independent experts and comparisons between them are very rare and usually not published!

Building a Geodemographic System from Scratch

orWhy coming to THIS Seminar could be very VALUABLE and save you lots of money!

Step 1. Select some census data for the smallest areas (census enumeration districts)

Step 2. Refine choice the choice of variables

Step 3. Classify the census Ed's using a multivariate classifier

Step 4. Evaluate the Classification and decide whether more or less clusters are needed

Step 5. Invent good names for the clusters that will enhance their marketing appeal as well as provide instantly recognizable thumbnail sketches of the types of residential areas involved

Step 6. Decide on whether you want higher order groupings of the clusters

Step 7. Test it out in secret in case it doesn’t work!

Step 8. Create cluster-postcode directory

Step 9. Repeat Steps 1 to 8 for different market segments

Step 10. Repeat everything using a mixture of census and non-census data sources at the postcode level

Where does GIS come in?

Where does GIS come in?

It does have some limitations as well as strengths

its still BIG business

Geodemographics is a VERY SIMPLE Geo-Targeting Technology

In the 1990’s S4 became R4

So Why bother with Geodemographics at all?

Maybe Geodemographics can beimproved?

Lets Review the Key Features of Geodemographics..

“Ermm!!”YES

Geodemographics is really only a crude form of GeoTargetting

“SMARTGeodemographics?????????????????????”

There are TWO major problems here:

ItsReally a Modeling process!

So why bother Developing Geodemographics further?

Geodemographics needs to be re-developed as a 2001 technology!

Improving Geodemographics

However.. Better Classifiers seldom produce significantly better results!

It seems that the world’s BEST geodemographic classification with 50 clusters in it will nearly always be outperformed by a conventionally produced system with 75 or more clusters in it!

There is NO WAY of KNOWING a good number of clusters without EXPERIMENTATION specific to EACH application

Improving the parts is not SUFFICIENT and a more fundamental, re-think is necessary!

Important to improve the Process

Incorporation of Explicit Neighborhood Effects

Fuzzy Geodemographics was first suggested over 10 YEARS AGO but so far there is not a single operational system anywhere in the world!

Fuzzy Geodemographic Systems

Fuzzy Geodemographics should now be a feasible technology

The real problem is that we ALSO need a fundamentally different and new way of APPLYING Geodemographics

Performance is more dependent on the number of clusters used than on the quality of the classificationBUT the optimal number of clusters depends on:

A Intelligent Geodemographic Targeting Machine (IGTM/1)

IGTM/1 attempts to find the best available geodemographic segmentation system for a specific application

IGTM/1comes with a limited assurance that it will be difficult to obtain a better geodemographic targeting result

The IGTM/1 approach has been developed into a more comprehensive Targeting Machine that could handle ANY targeting whether Geodemographics or ranked scores or model results

Personal Loans Case studies

Its a Mix and Match Problem!

Next step is to develop an optimal combiner that seeks an optimization of targeting performance by selecting the best bits from the available whole segmentation systems

Are WE still being REALLY Geographically DUMB right now?

Two aspects of Geography that are important here..

Using a Geographical Analysis Machine (GAM) to find clusters of unusually HIGH response

Another Example for a different mailing

What? You do not already do this??

You really do need to start building ..Intelligent Hybrid Computational GeoMarketing Systems instead of using legacy Geodemographic Systems

Author: Stan Openshaw, School of Geography, University of Leeds

Email: stan@geog.leeds.ac.uk

Home Page: http://www.geog.leeds.ac.uk/staff/s.openshaw/