SEO Marketing Research

SEO Marketing Research header image 2

Marketing Models and Properties of a Good Model

31 Comments · models

According to one of the pioneers of marketing models, John Little (1970):
? good models are hard to find;
? good parameterization is even harder;
? managers do not understand models;
? most models are incomplete.

There can be little doubt that Little’s statement is of even greater relevance if we deal with models that aim to describe and analyze international or cross-cultural marketing problems.

Types of marketing models and their purpose
Models of consumer behaviour and media planning usually assume that consumers are exposed to advertising in a variety of print media, broadcast programming and TV channels.

While this applies in countries such as the US, France and Japan, it may not apply to consumers in Africa and the Middle East, either because the consumers are not literate, they cannot afford to approach such media, the media do not exist, or the government controls advertising in the media.

Models of product innovation or product launches are of little value if the market deems the product unacceptable.

An alcoholic beverage cannot be launched in Muslim countries, chewing gum has been prohibited in Singapore and there are increasing global restrictions on cigarette smoking.

Pricing models may apply to markets without government regulation of prices. But they are probably invalid or of limited validity when analyzing markets where the authorities have established minimum or maximum prices for goods and services.

Furthermore, can the same model be used for describing the German beer market, where more than 1,000 producers compete and the biggest competitor has a national share of 8 percent, and the Danish market, where one company, Carlsberg/T’uborg, has more than 70 percent of the market?

Distribution models assume that the appropriate vertical and horizontal channels are available, though there are variations, “home parties” (the key to Tupperware’s success) and “snowball systems” are legal ways of distributing goods in some countries, but forbidden or strictly regulated in other countries.

Availability of valid measurements constitutes a necessary but not sufficient condition of a good model.

As the table below shows, many other factors are also of importance.

? A model should be comprehensive and not omit any critical relationships, and yet be as simple as possible.

? All relationships that do not facilitate the understanding of the phenomenon in a significant way should be excluded.

? The model should be without error. This implies that external constructs or variables should not retard an understanding of internal causal links. Likewise, spurious effects should not be confounded with true relationships.

Properties of a good model

1. Comprehensive  8. Operational
2. Simple   9. Generalizable
3. Little “noise” or error  10. Communicable
4. Measurable  11. Has implications for managers
5. Valid and reliable  12. Successful
6. Robust   13. True
7. Logically consistent

? The variables of the model ought to be quantitatively measurable. While verbal, or qualitative, models may make sense, they are beyond the scope of this text.

? The overall model as well as the involved measurements must be both valid and reliable. Scales must work well and estimates make sense.

? It must be robust (stable and insensitive to minor changes in the environment). A model is said to possess robustness if it is difficult for a user to obtain incorrect answers.

? Consistency is needed. For example, since negative sales and prices make little sense in the real world, the model counterpart should satisfy the same constraints.

? Operational properties are important. While game theory and linear programming possess excellent theoretical properties they often lack everyday applicability.

? Hypothesized patterns and estimates ought to be generalizable. They should work properly in different settings.

However, a model of individual buying behaviour may work well in North America while it performs poorly in developing countries where many of the household’s purchases are physically carried out by the grandmother.

If this is the case, a refined model of family or household buying behaviour might better fit the data.

? Findings must be communicable. This implies that findings must facilitate an understanding of the problem to managers. It must make sense.

? Management must be able to transform the insight gained to tactical and/or strategic decisions with regard to a product launch, pricing, advertising, distribution and so on.

? The model must be successful. Unfortunately, a model’s long-term successfulness can only be assessed properly by posterity’s observers.

? A model has to be true as demonstrated by it agreeing with known facts.

The market for chewing gum in an industrialized country has been growing for decades and appears to keep expanding, but the population is growing slowly.

Closer inspection of the country’s population pyramid might raise concern about future sales.

Assume that birth rates have been declining for years and that the average age of the population is going up.

Why should this make us worry? Well, because the consumption of chewing gum is inversely related to age.

So within a few years, sales will probably start declining since the number of heavy users (teenagers) is declining while the number of elderly non-users is increasing.

Keywords: model, marketing models, variables,


31 Comments so far ↓

Leave a Comment