Premium
SIMPLE MODELS FOR MONITORING NEW PRODUCT PERFORMANCE
Author(s) -
Ayal Igal
Publication year - 1975
Publication title -
decision sciences
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.238
H-Index - 108
eISSN - 1540-5915
pISSN - 0011-7315
DOI - 10.1111/j.1540-5915.1975.tb01015.x
Subject(s) - aggregate (composite) , product (mathematics) , computer science , customer base , marketing , marketing mix , test (biology) , econometrics , panel data , product category , aggregate data , simple (philosophy) , consumer behaviour , advertising , economics , business , mathematics , statistics , paleontology , philosophy , materials science , geometry , epistemology , composite material , biology
Several studies have shown that consumers trying a new brand of frequently purchased consumer goods behave differently from repeat buyers. The “customer mix” for a new brand changes over time. Early in a new brand's life, most of the buyers are triers, while later in its life, a significant portion of the buyers are repeaters. The study presented here tests the hypothesis that an aggregate sales forecast, “averaging” the behaviors of triers and repeaters, would be outperformed by a model that treats them separately. A new product model, giving separate consideration to triers and repeaters, is developed. This model is tested on predictive ability against a widely used single‐equation aggregate model. Consumer panel purchase diaries and data on advertising in measured media, covering a period of 42 months following introduction of a major brand of cold tablets, serve as the proving grounds. Data for the early part of this period serve as the data base for both models, while data for the rest of the period serve to test the predictive ability of both models. The disaggregate model is shown to perform significantly better than the aggregate model in terms of predictive accuracy. It also offers several other advantages in use as a decision aid, both for GO‐NO decisions and for marketing mix decisions. Finally, several problems in the implementation of the proposed model and implications for research strategy are discussed.