Academic research occasionally desires to be practical. Various journals therefore often have sections that allow for the results of practical work with managers where there aren’t significant theoretical contributions. In the past the Marketing Science journal has published work under “Applications”. In this vein Doyle and Sanders did some work on multiproduct advertising budgeting.
The challenge they looked at was how to optimize advertising across a broad range of products. I.e. how to get the best bang for a set amount of advertising bucks given there are a large number of products the advertising budget could support. Doing this they worked with “a large European variety store chain” (Doyle and Sanders, 1990, page 97).
There are some interesting features of the context — such as more significant lags between advertising and its effects than we might have expected to see. This is because the customer sees the product in the catalogue and “customers often retain catalogues and respond several weeks after their distribution” (Doyle and Sanders, 1990, page 106).
The cross-elasticities are also interesting. This is just a fancy way of saying that advertising one product line leads to increased sales of another (unadvertised) product line. This is pretty standard in retail — some headline (traffic driver) products drag customers into the store where they buy other products too.
The method they use is essentially an empirical case study. They looked at how advertising could be improved at the retailer. There is, however, little theory, or even a theoretical discussion. They note that their model showed how advertising effectiveness could be improved at that retailer. They suggested less widespread advertising of products and “a much greater focus on products with high marginal response rates — notably childrenswear and candy” (Doyle and Sanders, 1990, page 107). The good thing about this research is that the relevance, i.e. application, is pretty obvious. The challenge is the question of how far the findings can be extended as it isn’t clear why the findings should apply anywhere beyond the single firm.
Still I think there is value in this sort of work. I wouldn’t want all, or even most, academic research to be this sort of work but I suspect it has great value to anchor academic research in the real world.
Read: Peter Doyle and John Sanders, 1990, Applications: Multiproduct Advertising Budgeting, Marketing Science, 9 (2), 97-113.