A number of different sales forecasting systems have been developed for predicting the sales volume potential of new products, particularly in the FMCG area. While most of these systems have a reliable track record for more conventional new products, they do not tend to be as good for innovative, image based products. This paper identifies some of the reasons why image based products are more difficult to deal with than most others. In particular it gives an example of a recent research project conducted by the authors in which a sales forecasting model was developed to this end. The model includes an application of catastrophe theory, a mathematical technique which helps explain discontinuities in behaviour.