DATA SOURCES AND THEIR GEOGRAPHICAL INTEGRATION
Geodemographics has come into use as a shorthand label for both the development and the application of area typologies that have proved to be powerful discriminators of consumer behaviour and aids to “market analysis”.
USES OF GEODEMOGRAPHICS
Naturally enough, many extravagant claims have been made about the capabilities of geodemographics. A recent brochure produced by CACI in the UK identifies nine different “applications of the ACORN Classifications for your Business”:
- Site analysis
- Media buying
- Direct Mail
- Sales planning
- Database Analysis
- Planning for Public Services
- Market Research Sample Frames
- Door-to-Door Leaflet Campagins (CACI 1993a)
For the analysis in this section, we will group these activities into three broader categories: market research; database marketing; and retail analysis. Let us consider each in turn.
- Market research
The key capability which is offered by any geodemographics system is the facility to apply a single identifying characteristic to any neighbourhood. One of the factors which makes geodemographics seductive is that the social geographies which are painted in this way are almost invariably extremely plausible. Potential clients might be invited to provide their home postcode, and often will be surprised to see a computer come back with a fairly accurate assessment of the type of neighbourhood in which they live.
One of the main market research applications of geodemographics comes if a client is able to produce a list of customer addresses. Each of these addresses can be postcoded, and each postcode can be assigned to a census ED. Each customer can then be ascribed the geodemographic label for the ED in which he or she resides. Once this exercise has been repeated for all of the customers in the database, it is then possible to build up a profile of the customer base in geodemographic terms.
A slightly different type of market research application for geodemographics is in the design of sample frames. A typical problem here is how to select stratified samples for a survey which are fully representative of the population.
- Database marketing
Given the ability to produce geodemographic profiles for particular brands and products, as described above, then we have a potentially powerful weapon for consumer targeting. Consider the example that 3.9 per cent of a population live in neighbourhoods of ACORN Type B03 – “Cheap modern private housing”. However, between them these people accounted for 6.4% of holiday camp visits. Suppose now that a company wishes to organize a direct mailing campaign for a new or existing holiday camp. It is likely that by focusing the campaign for a new or existing ACORN Type B03, it has a much higher chance of reaching potential customers than by targeting the population at random. It is this kind of process which forms the basis for using geodemographics within database marketing.
We can formalize the benefits of this type of procedure through some relatively straightforward analysis. For each ACORN group or type, we can form a “penetration index” by comparing the target and base populations within a category. In the above example, this goves us a penetration index of (6.4/3.9) x 100 = 161 for holiday camp visits within ACORN Type B03. At the other extreme, an index of only 20 is achieved within type I32 (“furnished flats, mostly single people”). These penetration indices are shown in the column labelled “Index” in the following table.
Table. Profile of households which use holiday camps
Next, we can rank the geodemographics groups in descending order on the penetration index. By plotting out the cumulative target and base populations we are able to form a “Lorenz curve” for the product in question. With a little thought, we can appreciate that the Lorenz curve will always be convex and above the line of equality (because we always start with an index greater than 1 for the top-ranking target group, and because the gradient of the curve is equal to the index). We can also see that the steeper the curve, the greater the level of “discrimination” that is being provided. The steepness of the curve can be measured by comparing the area under the curve with the area under the line of equality to give an index between 1 (zero discrimination) and 2 (perfect discrimination). This is generally known as the “Target Effectiveness Index”. Finally, note that for this reason the Lorenz curve of the following is sometimes known simply as “Gains Chart”.
Figure: A Lorenz curve for holiday camp visits
- Retail analysis
The most straightforward applications of geodemographics to retail analysis are intimately connected to GIS. We have seen from previous sections that geodemographic classifications allow us to attach a neighbourhood type to each small area in the country, such as a unit postcode or enumeration district. Suppose that we could create a coverage within a GIS showing the spatial extent, population and neighbourhood type of each area. If we were to overlay upon this coverage a set of retail locations, then basic GIS buffering operations upon this coverage a set of retail locations, then basic GIS buffering operations could be used to calculate the population of each demographic group within specified distance bands around each outlet, for example 1 mile, 2 miles, and so on. However, one obvious problem with this “banding” methodology is that straight-line distances do not reflect geographical accessibility very meaningfully. Thus if we are considering the retail potential at a particular location, it makes far more sense to consider drive time bands around an outlet or centre in preference to straight-line distances. Indeed products such as CCN’s “Environ” package comprise simple population aggregations within 15, 30 and 45-minute drive time bands around major retail centres in the country.
Next, a marginally more convincing approach is to combine drive time analysis with geodemographics.