Abstract:

The purpose of this paper is to provide a practitioner’s view of how TV reach and frequency estimates are developed in the United States of America. The paper is in four parts. The first part discusses the importance of reach and frequency as tools in the media planning process. One method to evaluate TV reach and frequency is the direct inquiry mode where the planner inputs a specific buy into the national ratings panel. This approach will not work in the United States, given the bulk buying process where most TV programs are not known in advance. The second section of the paper discusses the specific nature of a transformed linear regression approach used in the U.S. TV schedules are produced from the ratings service's panel. Regression analyses are performed against the schedules to produce reach estimates by demographic group. This approach has the virtue of being actionable and accurate. Its drawbacks are that reach levels can be skewed due to outliers. The paper then discusses how daypart reach levels can be combined on a random duplication basis. This utilizes an approach developed by Mark Maiville that adjusts the random duplication factor according to the average of the combined reach of the two schedules. This produces highly accurate estimates, although ad hoc corrections may need to be applied in any computer based application. The final section outlines how frequency distributions can be developed utilizing Metheringham’s variation on the Beta-Binomial. The method is accurate, however, the resulting smooth curve does not mirror the peaks and troughs of actual frequency distributions. This three tiered approach to reach and frequency, while accurate, is inelegant and contains several assumptions. New research services suggest the development of more unified and sophisticated approaches.