A Media Planning Calculus

Abstract

A convenient on-line computer system selects and schedules advertising media. The system consists of a market-response model, a heuristic search routine, and a conversational input-output program. The user supplies a list of media options, a budget, and various objective and subjective data about the media options and the desired audience. The system selects a set of options and schedules them over time, seeking to maximize total market response. The model of market response works as follows: The population is divided into market segments. People in each segment are characterized by their sales potential and media habits. Ads placed in the media options lead people to be exposed to the advertising. The pattern of exposures in each market segment is determined by media coverage and duplication data. People tend to forget exposures and so the retained exposure level decays in the absence of new advertising. The response of an individual, in terms of the fraction of sales potential realized by the advertiser, increases with exposure level but with diminishing returns. Total market response is a sum over people, market segments, and time periods. The calculation of response is based on analytic formulas that are computationally very efficient. A maximum-seeking, heuristic calculation starts with any schedule, adds options with a high increment of response per dollar and deletes options with a low increment per dollar until no more improvement can be found for the given budget. An on-line system, called mediac, permits the use of model and heuristic at a remote console of a time-shared computer. Communication with the computer is conversational and largely self-explanatory. The system is operational. Computing costs have been a fraction of a per cent of the cost of the media scheduled. Improvements over previous schedules, as calculated by the model from the user's input data, have run from 5 per cent to over 20 per cent.