Abstract:

This paper will describe the work that has been carried out in support of a modelling approach aimed at predicting the overall performance of a TV programme series based on the audience data for the first transmission(s) of that series. It is well known that a number of measures of viewing behaviour are related to the actual long-term ratings performance of programmes or series. For example, there is a great deal of anecdotal evidence demonstrating that a high Appreciation Index (AI) tends to lead to a growth in ratings (and vice versa). Previous work carried out by the authors also suggested that Programme Loyalty (the average proportion of a programme which viewers to that programme actually watched) and Series Loyalty (the levels of viewing duplication across episodes) could be very important discriminators in determining the longer term performance of a programme or series. The aim of this paper was to investigate the relationship between a number of different viewing measures which could be potential discriminators (applied to the earhest programmes in a series) and the overall audience performance of that series. The objective was to construct a theoretical model which describes viewing behaviour over the course of a programme series so that it would be possible to accurately predict the overall performance of that series based on the data available for the first one or two episodes. Any such model would be tested for both reliability and validity - it would have to predict to an acceptable level of accuracy and it would need to produce predictions of reasonably similar levels of accuracy for a range of different programme series.