Media Planning

How to Analyze Media Investments

Here are five ways to analyze how much to invest in various media.

1. Last year's advertising expenditures for the brand
   • Examine last year's expenditure and then list the arguments favoring increases along with those favoring decreases.
   • This should be done independently by two or more analysts who would then make a subjective estimate of whether to increase or decrease expenditures and by what percentage.
   • The different estimates would then be averaged.
   • Might also use the Delphi procedure here
2. Typical expenditures for the firms in your industry
   • The typical expenditure menthod assumes that your competitors are rational. Before looking at the typical expenditures, ask experts to write down all arguments as to whether you should be advertising more or less than what is typical.
   • Then ask them to provide estimates of how much your advertising should depart, in percentage terms, from the typical expenditure.
   • Estimates from different experts should then be averaged.
   • Consider using the Delphi procedure here
   • Then obtain information on typical expenditures for your industry. For more information on typical advertising expensidures, look at these sources.
3. Elasticity
   • Use the elasticity method in cases where you have an existing product for which you have been advertising.
4. Decomposition
   • Make estimates for each element of the decision to advertise
   • Use the decomposition method for any situation (e.g., for new products or existing products). The inputs for the method can come from judgement or from the data.
5. Experiment
   • The ideal way to allocate funds to advertising is to experiment.
   • This can be easily done for direct responce advertising. For example, if the advertising will be done by direct mail, try a sample of 10,000 names.
   • Examine the profit made from the resulting sales and divide by the costs of the direct mail ads to determine the return on investment.

References

Assmus, Gert, John U. Farley and Donald R. Lehmann (1984), "How advertising affects sales: Meta-analysis of econometric results," Jounal of Marketing Research, 21, 65-74.

Beltramini, Richard F. and Steven P. Brown (1994), "Miscomprehension a believability of information presented in print advertising," Advances in Consumer Research, 21, 218-223.

Sethuraman, Raj and Gerald J. Tellis (1991), "An analysis of the tradeoff between advertising and price discounting," Journal of Marketing Research, 28, 160-174.


Finding Typical Expenditures by Industry

• AdAge.com provides advertising to sales ratios for the 200 largest ad spending industries for the years 1997 - 2001. It gives the name of the industry, SIC number, ad dollars as percentage of sales, ad dollars as percentage of margin, and annual ad growth rate percentages.
• adinsite.com provides advertising-to-sales ratios for a few products, retail stores, and services. It gives average ad dollars as percent of sales for industry groups and sometimes for sales breakdowns within group. The year is 1995 from these sources: NAA Planbook; Schonfeld & Associates, Advertising Ratios and Budgets 1995; National Association of Retail Dealers of America, 1995.
• COMPUSTAT has annual advertising expenditures for many US public companies (limited to University of Pennsylvania users)
  1. Choose Compustat
  2. From Compustat menu, choose Industrial Annual Current.
  3. Choose date
  4. Screen by ticket, CUSP or SIC (they call it DNUM)
  5. For display choose company names or additional variables
  6. Choose Data12 for Sales
  7. Choose Data 45 for Advertising expense
  8. Display report
  9. Click on "Advanced Query" to screen on other variables, e.g. assets size, or to download as excel file

Using the Elasticity Method

One way to estimate the effectiveness of an ad is to first assume that an ad has a typical elasticity of about +0.11. This is based on the Sethuraman and Tellis paper from the Journal of Marketing Research, 1991, pp. 160-174). Better yet, averages for various ads or products or markets can be selected from the Setheramun and Tellis paper, Table 1. Next, it would be adjusted based on its ESAP score. A brief summary is provided here.



Considerations from Assmus, Farley and Lehman (1984):

• Elasticities are lower at the brand level than at the product level.
• Elasticities are higher during the introductory phase of a product.

Elasticities should be higher for ads with a high ESAP score.



Here is an example based on the Milk Mustache ad campaign in the mid-1990:

Find out the current level of advertising (say, over the past year). For milk, assume that it was $52 million. Calculate expected cost of new campaign (estimate it as $52 million so that the increase is 100%. Calculate industry sales (Time magazine said it was $14 billion). The campaign is for a nondurable (.09) in the US (.09) using print media (.13), so an overall estimate of .11 seems reasonable. It is at the product level, so add .01 to get .12. The ad had a somewhat low average score on the ESAP, so deduct .02, giving an overall estimate of .10.

Increase in revenue from a 100% increase in advertising, then, would be 10%. 10% of the $14 billion is $1.4 billion dollars increase. Assume that profit is 2% of sales. Thus, increased profit would be $28 million. Given these assumptions, and estimates, the return would not cover the $52 million for the ad campaign.

Recent Update: About a year after this analysis was completed, the AMA Marketing News reported that the Milk Mustache campaign involved an expenditure of $190 million (during 1997-8), and that it won advertising awards. However, milk consumption decreased in 1997. There had been no forecast of what would have happened to milk consumption had the marketers not run the advertising campaign. However, data on milk consumption in the U.S. from 1988 to 1996 revealed that consumption was relatively constant overall, but that it fluctuated by as much as 1.8% from year to year. A linear extrapolation would show no trend.


Using the Decomposition Method

Decomposition is especially useful when the breakdown allows one to draw upon different sources of knowledge and when none of the components is subject to high uncertainty (compared with the uncertainty in the global estimate of elasticity). For example, to estimate “understanding,” one might draw upon Beltramini and Brown (1994). They estimated correct comprehension of 80% for print ads and 73% for TV commercials.

Given reliability problems, it is suggested that you obtain independent estimates from at least five people, each of whom has knowledge about the effect of advertising.

Ask each expert to provide an estimate of the expected effect of the ads, and also 95% confidence intervals about the estimates. Summarize the median for each estimate and calculate a typical confidence interval.

The decomposition procedure will guide you through this process. Here is the format for a print ad.



A note on the above: Return on Investment should measure only new costs, not those already spent.

If you would like to view this page as a separate html file, Click Here